Full download ebooks at ebookmeta.

com

Sliding Mode Control of Semi-Markovian Jump

Systems 1st Edition Jiang

https://ebookmeta.com/product/sliding-mode-control-of-semi-
markovian-jump-systems-1st-edition-jiang/

OR CLICK BUTTON

DOWLOAD NOW

Download more ebook from https://ebookmeta.com

More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Synchronization Control of Markovian Complex Neural

Networks with Time-varying Delays 1st Edition Junyi
Wang

https://ebookmeta.com/product/synchronization-control-of-
markovian-complex-neural-networks-with-time-varying-delays-1st-
edition-junyi-wang/

Fuzzy System Identification and Adaptive Control

Communications and Control Engineering Ruiyun Qi Gang
Tao Bin Jiang

https://ebookmeta.com/product/fuzzy-system-identification-and-
adaptive-control-communications-and-control-engineering-ruiyun-
qi-gang-tao-bin-jiang/

Microbiologically Influenced Corrosion of Concrete

Sewers Mechanisms Measurements Modelling and Control
Strategies 1st Edition Guangming Jiang Ed

https://ebookmeta.com/product/microbiologically-influenced-
corrosion-of-concrete-sewers-mechanisms-measurements-modelling-
and-control-strategies-1st-edition-guangming-jiang-ed/

Analysis and Synthesis of Singular Systems Emerging

Methodologies and Applications in Modelling
Identification and Control 1st Edition Zhiguang Feng
Jiangrong Li Peng Shi Haiping Du Zhengyi Jiang
https://ebookmeta.com/product/analysis-and-synthesis-of-singular-
systems-emerging-methodologies-and-applications-in-modelling-
identification-and-control-1st-edition-zhiguang-feng-jiangrong-
Encyclopedia of Systems and Control John Baillieul
(Editor)

https://ebookmeta.com/product/encyclopedia-of-systems-and-
control-john-baillieul-editor/

Control Systems 1st Edition Ashfaq Husain Haroon Ashfaq

https://ebookmeta.com/product/control-systems-1st-edition-ashfaq-
husain-haroon-ashfaq/

Modeling and Control of Drug Delivery Systems 1st

Edition Ahmad Taher Azar

https://ebookmeta.com/product/modeling-and-control-of-drug-
delivery-systems-1st-edition-ahmad-taher-azar/

The Dynamic Morse Theory of Control Systems 1st Edition

Josiney A Souza

https://ebookmeta.com/product/the-dynamic-morse-theory-of-
control-systems-1st-edition-josiney-a-souza/

Kissing Mr Wrong Mercy Indiana 1 1st Edition Shirley

Jump Jump Shirley

https://ebookmeta.com/product/kissing-mr-wrong-mercy-
indiana-1-1st-edition-shirley-jump-jump-shirley/
Sliding Mode Control
of Semi-Markovian
Jump ­Systems
Sliding Mode Control
of Semi-Markovian
Jump ­Systems

Baoping Jiang and Hamid Reza Karimi

MATLAB® is a trademark of The MathWorks, Inc. and is used with permission. The MathWorks
does not warrant the accuracy of the text or exercises in this book. This book’s use or discussion
of MATLAB® software or related products does not constitute endorsement or sponsorship by The
MathWorks of a particular pedagogical approach or particular use of the MATLAB® software

First edition published 2022

by CRC Press
6000 Broken Sound Parkway NW, Suite 300, Boca Raton, FL 33487-2742

and by CRC Press

2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN

© 2022 Baoping Jiang and Hamid Reza Karimi

CRC Press is an imprint of Taylor & Francis Group, LLC

Reasonable efforts have been made to publish reliable data and information, but the author and
­publisher cannot assume responsibility for the validity of all materials or the consequences of
their use. The authors and publishers have attempted to trace the copyright holders of all material
­reproduced in this publication and apologize to copyright holders if permission to publish in this
form has not been obtained. If any copyright material has not been acknowledged please write and
let us know so we may rectify in any future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known
or hereafter invented, including photocopying, microfilming, and recording, or in any information
storage or retrieval system, without written permission from the publishers.

For permission to photocopy or use material electronically from this work, access www.copyright.com
or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923,
978-750-8400. For works that are not available on CCC please contact m ­ pkbookspermissions@
tandf.co.uk

Trademark notice: Product or corporate names may be trademarks or registered trademarks and are
used only for identification and explanation without intent to infringe.

ISBN: 978-0-367-56503-9 (hbk)

ISBN: 978-0-367-56504-6 (pbk)
ISBN: 978-1-003-09804-1 (ebk)

Typeset in Times
by codeMantra
Contents
Preface.......................................................................................................................ix
Acknowledgment.......................................................................................................xi
Authors.................................................................................................................... xiii

Chapter 1 Introduction...........................................................................................1
1.1  Sliding Mode Control.................................................................1
1.1.1 Basic Concepts of SMC.................................................1
1.1.2 Implementation of SMC................................................4
1.1.2.1 Sliding Surface Design..................................5
1.1.2.2 Sliding Mode Controller Design....................5
1.2 Semi-Markovian Jump Systems.................................................5
1.2.1 Review of Markovian Jump Systems............................5
1.2.2 Description of Semi-Markovian Jump Systems............8
1.3 Preview of This Book............................................................... 11
1.4 Some Useful Definitions and Lemmas..................................... 13
1.5 Abbreviations and Notations.................................................... 14
References........................................................................................... 15

Chapter 2 Stochastic Stability of Semi-Markovian Jump Systems

with Generally Uncertain Transition Rates......................................... 21
2.1 Introduction.............................................................................. 21
2.2 System Description................................................................... 22
2.3 Stochastic Stability Analysis....................................................26
2.4 Numerical Examples................................................................34
2.5 Conclusion................................................................................ 36
References........................................................................................... 37

Chapter 3 Fuzzy Integral Sliding Mode Control of Semi-Markovian

Jump Systems...................................................................................... 39
3.1 Introduction.............................................................................. 39
3.2 System Description...................................................................40
3.3 Main Results............................................................................. 42
3.3.1 Sliding Surface Design................................................ 43
3.3.2 Stochastic Stability Analysis....................................... 45
3.3.4 Reachability of Sliding Surface.................................. 52
3.4 Numerical Example.................................................................. 55
3.5 Conclusion................................................................................ 61
References........................................................................................... 61

v
vi Contents

Chapter 4 Fuzzy Sliding Mode Control for Finite-Time Synthesis of

Semi-Markovian Jump Systems.......................................................... 65
4.1 Introduction.............................................................................. 65
4.2 System Description...................................................................66
4.3 Main Results............................................................................. 69
4.3.1 Finite-Time Reachability of Sliding Surface in T*...... 69
4.3.2 Finite-Time Boundedness Analysis During [0, T*]..... 71
4.3.3 Finite-Time Boundedness Analysis of Sliding
Mode Dynamics����������������������������������������������������������74
4.3.4 Finite-Time Boundedness Analysis Over [0,T]........... 76
4.4 Numerical Example.................................................................. 81
4.5 Conclusion................................................................................ 86
References........................................................................................... 86

Chapter 5 Adaptive Fuzzy Sliding Mode Control of Semi-Markovian

Jump Systems...................................................................................... 87
5.1 Introduction.............................................................................. 87
5.2 System Description................................................................... 88
5.3 Main Results............................................................................. 91
5.3.1 Sliding Mode Observer Design................................... 91
5.3.2 Design of Sliding Surface........................................... 93
5.3.3 Stochastic Stability and H-Infinity
Performance Analysis������������������������������������������������ 95
5.3.4 Reachability Analysis................................................ 104
5.4 Numerical Example................................................................ 106
5.5 Conclusion.............................................................................. 111
References......................................................................................... 112

Chapter 6 Decentralized Adaptive Sliding Mode Control of Large-Scale

Semi-Markovian Jump Systems........................................................ 115
6.1 Introduction............................................................................ 115
6.2 System Description................................................................. 116
6.3 Main Results........................................................................... 119
6.3.1 Sliding Surface Design.............................................. 119
6.3.2 Stability Analysis of Sliding Mode Dynamics.......... 119
6.3.3 Convergence of Sliding Surface................................ 124
6.4 Numerical Example................................................................ 127
6.5 Conclusion.............................................................................. 131
References......................................................................................... 131
Contents vii

Chapter 7 Reduced-Order Adaptive Sliding Mode Control for Switching

Semi-Markovian Jump Delayed Systems.......................................... 133
7.1 Introduction............................................................................ 133
7.2 System Description................................................................. 134
7.3 Main Results........................................................................... 136
7.3.1 SMC Law Synthesis.................................................. 136
7.3.2 Mean-Square Exponential Stability Analysis........... 137
7.3.3 Adaptive SMC Law Design....................................... 145
7.4 Numerical Examples.............................................................. 147
7.5 Conclusion.............................................................................. 155
References......................................................................................... 155

Outlook................................................................................................................... 157
Index....................................................................................................................... 159
Preface
Recently, semi-Markovian jump systems (S-MJSs) have received wide attention due
to their feasibility in modeling practical systems, for instance, electrical systems,
economics and mechanics, that suffer from abrupt structural changes caused by phe-
nomena such as random failures or repairs, changes in subsystem interconnections
and so on. Nowadays, the analysis and synthesis of S-MJSs is becoming more and
more rich, and many important results have been achieved. However, due to the gen-
erality that the transition rates in S-MJSs are time-varying and hard to obtain in prac-
tice, the fundamental issues of stochastic stability and stabilization of such systems
are still challenging. Hence, we make an attempt to tackle the analysis and synthesis
of a class of continuous-time S-MJSs with generally uncertain transition rates.
Sliding mode control has unique advantages in dealing with nonlinear complex
systems: the sliding mode control model has the properties such as fast response and
good transient performance, strong robustness to system perturbations and uncertain-
ties, which has attracted great attention in the control community since its appear-
ance. For sliding mode control of stochastic systems, the two-step design become
more complicated than normal state-space systems; it is necessary to consider the
jumping effect caused by the switching modes, the finite-time reachability due to
jumping rules, the controller design with deficiency-mode transition rates informa-
tion, etc. This monograph contains valuable references and knowledge to help the
relevant researchers to explore these issues and carry out further research in the area.
The purposes of this monograph are to present latest development and literature
review on sliding mode control of S-MJSs, which involves problems such as sto-
chastic stability analysis, fuzzy integral sliding mode control, finite-time sliding
mode control, adaptive sliding mode control and decentralized sliding mode control
for S-MJSs, and their applications in robotic manipulator and circuit systems, for
instance. The contents are also suitable for a one-semester graduate course.
More specially, in this monograph, basic concepts and results on stochastic sta-
bility of S-MJSs are first presented in Chapter 1. In Chapter 2, based on the former
fundamental results, the problem of stochastic stability for S-MJSs with generally
uncertain transition rates (TRs) is investigated through linear matrix inequality
(LMI) technique. Chapter 3 deals with the issue of robust fuzzy integral sliding
mode control for continuous-time Takagi–Sugeno fuzzy model-based S-MJSs, and
the assumption that input matrices are the same with full column rank is removed.
Chapter 4 treats finite-time sliding mode control of continuous-time S-MJs with
immeasurable premise variables via fuzzy approach, and LMI conditions are proposed
to guarantee boundedness performance both at the reaching phase and at the sliding
motion phase. In Chapter 5, observer-based adaptive sliding mode control for non-
linear Takagi–Sugeno fuzzy model-based S-MJSs with immeasurable premise vari-
ables is investigated, LMI conditions for stochastic stability with an H ∞ performance
disturbance attenuation level γ of the sliding mode dynamics and error dynamics are
developed, and an adaptive controller is synthesized to ensure finite-time reachability
of a predefined sliding surface. Chapter 6 presents a decentralized adaptive sliding

ix
x Preface

mode control scheme for the stabilization of large-scale semi-Markovian jump-inter-

connected systems, in which adaptive law is designed to compensate dead-zone input
nonlinearity and unknown interconnections. Chapter 7 gives reduced-order sliding
mode control approach to stabilize delayed-switching S-MJSs, LMI conditions are
proposed for mean-square exponential stability analysis, and adaptive controller is
designed for finite-time reachability purpose.

MATLAB® is a registered trademark of The MathWorks, Inc. For product

information,
please contact:
The MathWorks, Inc.
3 Apple Hill Drive
Natick, MA 01760-2098 USA
Tel: 508-647-7000
Fax: 508-647-7001
E-mail: info@mathworks.com
Web: www.mathworks.com
Baoping Jiang
Hamid Reza Karimi
Acknowledgment
This work is supported in part by The National Natural Science Foundation of China
under Grant 62003231; partially supported by The Natural Science Foundation
of Jiangsu Province under Grant BK20200989; Partially funded by the NCF for
colleges and universities in Jiangsu Province under Grant 20KJB120005 and the
China Postdoctoral Science Foundation (2021M692369), and partially supported
by the Italian Ministry of Education, University and Research through the Project
“Department of Excellence LIS4.0-Lightweight and Smart Structures for Industry 4.0”.

xi
Authors
Dr. Baoping Jiang
School of Electronic and Information Engineering, Suzhou
University of Science and Technology, Suzhou 215009, China
(e-mails: baopingj@163.com)
Baoping Jiang received the Ph.D. degree in control theory
from the Ocean University of China, Qingdao, China, in 2019.
From 2017 to 2019, he was a joint training Ph.D. candidate in
the Department of Mechanical Engineering, Politecnico di
Milano, Milan, Italy. In 2019, he joined the Suzhou University
of Science and Technology, Suzhou, China, where he is an associate professor. His
research interests include sliding mode control, stochastic ­systems, etc.

Dr. Hamid Reza Karimi

Department of Mechanical Engineering, Politecnico di
Milano, 20156 Milan, Italy, Email: hamidreza.karimi@
polimi.it
Hamid Reza Karimi received the B.Sc. (First Hons.)
degree in power systems from the Sharif University of
Technology, Tehran, Iran, in 1998, and the M.Sc. and
Ph.D. (First Hons.) degrees in control systems engineering
from the University of Tehran, Tehran, in 2001 and 2005,
respectively. He is currently Professor of Applied Mechanics in the Department of
Mechanical Engineering, Politecnico di Milano, Milan, Italy. His current research
interests include control systems and mechatronics with applications to automotive
control systems, robotics, vibration systems and wind energy.
Dr. Karimi is currently serving as the Chief Editor, Technical Editor and Associate
Editor for some international journals. He is a Fellow of The International Society
for Condition Monitoring (ISCM), a member of The Agder Academy of Science
and Letters and also a member of the IEEE Technical Committee on Systems
with Uncertainty, the Committee on Industrial Cyber-Physical Systems, the IFAC
Technical Committee on Mechatronic Systems, the Committee on Robust Control
and the Committee on Automotive Control. Dr. Karimi has been awarded as the
2016–2020 Web of Science Highly Cited Researcher in Engineering, the 2020
IEEE Transactions on Circuits and Systems Guillemin-Cauer Best Paper Award,
August-Wilhelm-Scheer Visiting Professorship Award, JSPS (Japan Society for
the Promotion of Science) Research Award, and Alexander-von-Humboldt-Stiftung
Research Award, for instance. He has also participated as General Chair, keynote/
plenary speaker, distinguished speaker or program chair for many international con-
ferences in the areas of control systems, robotics and mechatronics.

xiii
 1 Introduction

1.1 SLIDING MODE CONTROL

Sliding mode control (SMC) has received considerable attention since its first appear-
ance in the 1950s by Emelyanov and then was developed by Utkin, which has been
proven to be an effective robust control strategy for incompletely modeled or non-
linear systems [1–3]. Essentially, SMC is a special kind of nonlinear control, and its
nonlinearity lies in control of discontinuity, which makes SMC differ from other
control methods – that is, the structure of the controlled system remains unchanged
in the dynamic process, but can constantly change according to the current state
of the system (such as its deviation and derivative, etc.) to meet a desired require-
ment. The consequence is that the system-state trajectories will be forced onto a
specified manifold and kept a satisfactory sliding motion. Since the sliding mode can
be designed independent of the object parameters and disturbances, the SMC has
the advantages such as fast response, insensitivity to parameter changes and distur-
bances, without the need of on-line identification for the system, and simple physical
implementation. However, SMC has one main unavoidable defect: the sliding motion
is difficult to strictly remain to the equilibrium point after the state trajectory reaches
onto the sliding manifold, but to cross back and forth on both sides of the sliding
surface, which is the resource of chattering problem. Despite the disadvantage, the
researches and applications of SMC are rich, and the developments are continuing.

1.1.1 Basic Concepts of SMC

The continue-time case:
Variable structure control with sliding mode is called sliding mode variable structure
control or SMC. The system structure changes constantly according to switching
signals on both sides of S(x) = 0 in the state space. The principle of switching is the
called control strategy, which guarantees the existence of sliding mode dynamics.
Accordingly, S = S(x) and S(x) = 0 are called switching functions and switching sur-
faces, respectively. The sliding mode means the movement that the motion point
(state variable) of the system is attracted to the region as it approaches this region.
The motion of the system in the sliding mode region is called “sliding mode motion,”
which has the property that the motion has nothing to do with the plant changed
parameters and disturbances.
For a general nonlinear system of the form

x = f ( x , u, t ), x ∈  n , u ∈  m , t ∈ , (1.1)

where x (t ) is the system-state vector, and u(t ) is the control input.

1
2 Sliding Mode Control of Semi-Markovian Jump Systems

Now, given a switching function of the form S ( x ) = [s1 ( x ) s2 ( x )  sm ( x )]T , we

need to design a sliding mode controller u(t ) = [u1 (t ) u2 (t )  um (t )]T that

 ui+ (t ), si (t ) > 0,
ui (t ) =  −
u (t ), si (t ) < 0,
 i

in which ui+ ≠ ui− , such that the following three conditions are satisfied:

1. The sliding mode motion exists, that is, S ( x ) S ( x ) ≤ 0;

2. The reaching condition is satisfied; that is, all system trajectories outside
the sliding surface S ( x ) = 0 will reach onto the predefined sliding surface in
finite time subsequently maintained;
3. The dynamics in sliding surface S ( x ) = 0, that is, the sliding mode dynam-
ics, is stable with some specified performances.

The discrete-time case:

Discrete SMC is also known as SMC of discrete-time systems; that is, it extends the
SMC theory in continuous-time systems to discrete-time systems in order to meet
the requirements in current digital computer control systems. The study of discrete
SMC began in the 1980s [4], and its control strategies can be generally divided into
inequality arrival conditions and equality arrival conditions.
In the discrete-time case, the possibility that the system-state trajectories strictly
slide on the switching surface is almost zero. Therefore, the SMC of discrete-time
system comes up with the issue of “quasi-sliding mode”; that is for the discrete-time
systems, the SMC cannot produce an ideal sliding motion, but can only produce the
quasi-sliding motion, which also requires revisit of some basic problems in discrete
SMC, involving the existence of sliding mode, reachability and stability of sliding
motion
To establish SMC for the discrete-time system, its basic principle is exactly the
same as that of continuous-time system, which is also divided into two basic issues:
First: select a switching function S ( k ) to ensure that the sliding mode is globally
asymptotically stable;
Second: calculate the sliding mode controller u( k ) = u ± ( k ), so that all movements
can be reached onto the switching surface S ( k ) = 0 in a finite time.
As mentioned above, the arrival conditions in the discrete-time case are mostly
extended from the continuous-time case, and the following arrival conditions have
been proposed.
The inequality type:

[s( k + 1) − s( k )]s( k ) ≤ 0,

| s( k + 1) |≤| s( k ),

1 2
V ( k + 1) − V ( k ) ≤ 0, V ( k ) = s ( k ).
2
Introduction 3

The discrete-time exponential reaching law type:

s( k + 1) = (1 − qT )s( k ) − Tsgn(s( k )),

in which  > 0, q > 0 and qT < 1.

The discrete-time exponential reaching law condition differs from the general
inequality type of reaching condition in that the influence of sampling period is taken
into account. Therefore, there are several advantages for the discrete-time exponen-
tial reaching law:

• The system has good quality during the reaching phase, and the parameters
ϵ and q in the reaching law can be adjusted;
• The size of the switching band can be calculated;
• Solving the SMC problem becomes more simple;
• The reaching condition in the form of equation gives equation type of the
variable structure control, which is easy to impelment in the design process.

Even SMC has the advantages such as fast response, insensitivity to parameter varia-
tions and complete rejection of matched external disturbances, and shows strong
robustness to external noise interference and parameter perturbation. However, SMC
also shows its main disadvantage (chattering effect) in practical application due to the
hysteresis of the switching device; the actual sliding mode cannot always remain on
the switching manifold, which may be for the following reasons:

1. Space-lag switch
Switch lag corresponds to the existence of dead zone of state variables in the
state space. Therefore, the consequence is that constant amplitude wave will
be superimposed on the sliding surface.
2. Time-lag switch
In the vicinity of sliding surface, due to the time lag of the switch, the exact
change of the control effect on the state is delayed for a period of time, and
since the amplitude of the control quantity decreases gradually, it shows a
decaying triangular wave on the sliding surface.
3. The effect of physical inertia
Since the energy and acceleration of the system are finite, and the inertia of
the system always exits, the control switch is always accompanied by a time
lag, which is consistent with the time lag in effect.
4. Chattering generated by the characteristics of the discrete system.
The sliding mode of discrete-time system itself is a “quasi-sliding mode”; its
switching action does not occur on the sliding surface, but on the surface of
a conical body whose vertex is the origin point.

In addition, it is well known that the chattering cannot be eliminated, so we need

to select appropriate gain parameters to reduce the impact of chattering when
designing the sliding mode controller. At present, there are a lot of researches on
chattering phenomenon in SMC worldwide, and some conclusions are presented
4 Sliding Mode Control of Semi-Markovian Jump Systems

as follows: (1) quasi-sliding mode method mainly includes the continuous function
approximation method and boundary-layer design method; in order to smooth the
switching signal sgn(s(t)), a continuous sigmoid function was often proposed [5,6].
In Ref. [7], nonadaptive chattering-free sliding mode controllers were proposed
by assuming that the norm of the derivative of the sliding surface was upper-
bounded; (2) reaching law methods, such as using the linear approach law and
exponential approach law [8]; (3) the filtering method, which enables the control
signal of the system passing through the filter for smooth filtering; for instance, in
Ref. [9], a gain-scheduled SMC scheme was proposed for tracking control tasks
of multilink robotic manipulators, in which two classes of low-pass filters were
introduced to work concurrently for the purpose of acquiring equivalent control
in order to reduce the switching gains; (4) the interference observer method, using
the interference observer to estimate external perturbations and uncertainties,
then to compensate them [10,11]; (5) the dynamic sliding mode method – in order
to obtain the essentially time-continuous dynamic SMC law, the switching func-
tion used in the conventional SMC was reconstructed through a differential pro-
cess [12,13]; (6) fuzzy methods – one is based on experience, using fuzzy logic to
realize self-adjustment of SMC parameters, and the other way is based on the uni-
versal approximation property of fuzzy models [14,15]. Other methods to reduce
the chattering effect include the neural network, genetic algorithm optimization
and switching gain reduction [16–19].

1.1.2 Implementation of SMC
Generally, a conventional SMC design involves two steps: Step 1. Designing a
sliding surface S ( x ) = 0 such that the dynamics restricted to the sliding surface
has the desired properties, such as stability, disturbance rejection capability and
tracking; Step 2. Designing a discontinuous feedback controller u(t ) such that
the system-state trajectories can be attracted onto the designed sliding surface
in a finite time and maintained on the sliding surface for all subsequent time
periods. Figure 1.1 plots the implementation of an ideal SMC process, which
includes two phases – one is the reaching phase: from initial state x 0 to point A
on hyperplane – and the other is the sliding motion phase: from point A to the
equilibrium point O.

FIGURE 1.1 Implementation of an ideal SMC.

Introduction 5

1.1.2.1  Sliding Surface Design

The design of sliding surface is an important part of SMC, which determines the
dynamic quality of the obtained sliding mode to a large extent. For different SMC
strategies, different switching surfaces can be proposed, such as linear sliding sur-
faces [20], time-varying sliding surfaces [21], and integral sliding surfaces [22]. In
terminal SMC [23], a nonlinear function was introduced in the sliding surface design
to realize that the tracking error of the system in sliding mode converged to zero in
finite time. In global robust SMC [24], a class of dynamic switching functional is
designed to ensure that the system has dynamic sliding quality during the whole pro-
cess; that is, the reaching motion is eliminated and the system shows strong robust-
ness in the whole response process. The integral sliding surface method can reduce
steady-state error of the system and realize global robustness of the system. In view
of the aforementioned advantages, these sliding surface design schemes have been
widely used in control analysis of various systems.

1.1.2.2  Sliding Mode Controller Design

Another important part of SMC is the design of control law, that is, to find the appropri-
ate controller such that the system meets at least the three properties (the three points in
continue-time SMC). So far, with the extensive development of SMC theory, it has been
widely applied in the field of robotics and aerospace. The design of sliding mode control-
ler can also be roughly divided into the following categories: (1) the SMC of discrete-
time system [25,26] – the main problems to be considered are the chattering effect and
large gain feedback; (2) the SMC of mismatched uncertain system [27,28]; (3) the SMC
of time-delay systems [29,30]; (4) the SMC of nonlinear system [31,32]; (5) the SMC of
stochastic systems [33,34], and other types of complex systems. In terms of control meth-
ods, it is worth noting the following hot research fields: (1) the synthesis of adaptive SMC
strategy [35,36]; (2) the construction of the sliding mode observer [37,38]; (3) design of
fuzzy sliding mode controller [39,40]; (4) design of neural sliding mode controller [41,42];
and (5) the construction of high-order SMC [43,44]. Similarly, with the development of
control technology and computer technology, the back-stepping method [45], the linear
matrix inequality (LMI) technology [46] and the S-function method [47] have also rap-
idly promoted the development of SMC.

1.2 
SEMI-MARKOVIAN JUMP SYSTEMS
1.2.1 Review of Markovian Jump Systems
With the fast development of modern industrial engineering, the automatic control
theory has been applied, which roughly has gone through the following three stages:
classical control theory, modern control theory and large-system control theory. On the
contrary, the fast development of automatic control theory also promoted the devel-
opment of modern industrial engineering. In modern industrial engineering, the high
demands for modeling complex system with more accuracy and precision are increas-
ing. For example, some abrupt changes in the system, such as dynamic abrupt changes,
parameter shifting or environmental noise changes, need to be described by more
appropriate mathematical models. In general, these internal or external uncertainties
6 Sliding Mode Control of Semi-Markovian Jump Systems

FIGURE 1.2 RLC series circuit.

cannot be described by traditional state-space models, so experts and scholars intro-

duced stochastic processes and stochastic variables. Among the models, the application
of stochastic jumping system to model such a physical model with multimode property
or an intelligent control system with multicontroller switches has attracted great atten-
tion. Particularly, such kinds of stochastic system theory and stochastic control meth-
ods have been widely used in all walks of life, such as the network communications,
power systems and chemical engineering and many other engineering fields.
Consider an electrical circuit illustrated in Figure 1.2, in which the position of
the switch r (t ) changes randomly among the capacitors with three states  = {1,2,3}.
Accordingly, let x1 (t ) = uc (t ) and x 2 (t ) = iL (t ) as state vectors, where uc (t ) and iL (t )
are the voltage of the capacitor and the current of the inductor, respectively. So the
model in Figure 1.2 can be represented by the following dynamics:

diL (t )
L + uc (t ) + R1iL (t ) = u(t ),
dt

duc (t ) uc (t )
C (rt ) + = iL (t ).
dt R2

By defining x (t ) = [ x1T (t ) x 2T (t )]T , then it obtains the state-space system in the form:

 1 1 
 −   0 
 C ( r )R2 C ( rt )   
x(t ) =  t
 x (t ) +  1  u(t ).
 1 R1   L 
− −
 L L 
Introduction 7

The above state-space equation gives the form of a physical system with multimode
property modeled by a stochastic system.
As an important class of stochastic hybrid systems, Markovian jumping systems
(MJSs) have attracted the attention of many scholars due to their extensive modeling
applicability. Moreover, a large number of research results have been extended to
various social production processes, such as power system, manufacturing system,
communication system, control of nuclear power plant, control of aircraft and wire-
less servo control, etc. [48–51]. Since Krasovskii and Lidskii [52] proposed the MJSs
to describe the stochastic jumping system, a large number of scholars have given
many important theoretical achievements on the analysis of various performance
indexes and comprehensive control of such systems. These theoretical achievements
not only promoted the development of control theory, but also made great contribu-
tions to the improvement of social and economic benefits.

Definition 1.1 [53]

A random sequence {X (t , n ≥ 0), t ≥ 0} is a Markov chain if for all,

i0 , i1 , in , in+1 ∈ S , n ∈ N and Pr{X n+1 = in+1 | X 0 = i0 , X1 = i1 ,, X n = in} > 0, it satisfies

Pr { X n +1 = in +1 | X 0 = i0 , X1 = i1 ,, X n = in } = Pr { X n +1 = in +1 | X n = in } . (1.2)

Definition 1.2 [54]

A Markov chain {X (t , n ≥ 0), t ≥ 0} is homogeneous if the probabilities (1.2) do not

depend on n and is non-homogeneous in the other cases.
On the complete probability space (Ω, F , P ), consider the following linear MJS:

 x(t ) = A(rt ) x (t ) + B(rt )u(t )

 (1.3)
x (t ) = φ (t ),


where x (t ) is the system-state vector, A(rt ) and B(rt ) are the system matrices with
compatible dimensions that depend on rt . {rt , t ≥ 0} is a continuous-time homogeneous
Markov process taking values in a finite space  = {1,2,, s}. The evolution of the
Markov process {rt , t ≥ 0} is governed by the following probability transitions:

 π ij h + o(h), i ≠ j,
Pr{rt + h = j | rt = i} = 
 1 + π ii h + o(h), i = j,

where h > 0 and lim o(h) / h = 0 , π ij > 0, i ≠ j , is the transition rate from mode i at
h→ 0

time t to mode j at time t + h, and π ii = − ∑π

j ≠i
ij < 0 for all i ∈ .
8 Sliding Mode Control of Semi-Markovian Jump Systems

Definition 1.3 [55,56]

For the linear SMJ (1.3), for all initial conditions x 0 , r0 ,

I: if it satisfies that

 t

lim E 
t →+∞
 ∫  x(s) 
0
2
ds | x 0 , r0  < +∞,

then the system (1.3) is said to be stochastically stable;
II: if it satisfies that

{
lim E  x (t , x 0 , r0 ) 2 = 0,
t →+∞
}
then the system (1.3) is said to be mean-square asymptotically stable;
III: if for arbitrary parameters a > 0 and b > 0, it satisfies that

{ }
E  x (t , x 0 , r0 ) 2 < a  x 0 2 e − bt ,

then the system (1.3) is said to be mean-square exponentially stable.

IV: if it satisfies that

{
Pr lim E { x (t , x 0 , r0 ) } = 0 = 1,
t →+∞
}
then the system (1.3) is said to be almost surely (asymptotically) stable.

Lemma 1.1 [57]

The linear MJS (1.3) is stochastically stable if there exists positive-definite matrix Pi
such that the following condition is satisfied
s

Pi Ai + AiT Pi + ∑π P < 0.
j =1
ij j

1.2.2 Description of Semi-Markovian Jump Systems

It is worth noting that modeling stochastic jumping systems using Markov stochas-
tic processes or Markov chains has great advantages. However, in the current pace
of society and the speed of technological innovation, especially the rapid develop-
ment of computer technology combined with artificial intelligence, the needs on the
suitability of modeling a physical system are more rigorous. In traditional Markov
models, the transition rate induced by the Markov stochastic process is a constant,
which originates from the fact that the dwell time of its modes subjects to the mem-
oryless exponential distribution. However, the unique exponential distribution of
dwell time also greatly limits the effectiveness of MJSs in specific fields. As a result,
some of the theoretical and algorithmic research results for MJSs cannot be applied
to certain social industrial production effectively. Recently, the stochastic jumping
Introduction 9

system using semi-Markovian parameter has attracted the attention of many scholars
[58,59], because the semi-Markovian jump systems (S-MJSs) relax the restriction
that the distribution of mode dwell time obeys exponential distribution but obeys
more general distribution, such as the Gaussian distribution [60] and Weibull dis-
tribution [61], which further expands the analysis and synthesis of physical system.
From the point view of modeling stochastic systems, the MJSs and the S-MJSs have
great similarity that the theoretical research of the S-MJSs can borrow the results
and technical methods applied in MJSs for a certain extent. However, the transition
rate (TR) of an S-MJS is time-dependent, that is, time-varying. Therefore, the main
attention is focused on the influence of the time-varying TRs that play on the whole
system performance indexes and on the mechanism of control design. At present,
compared with the MJSs, there does not exist a set of mature and consistent theoreti-
cal basis for the study of the S-MJSs. This book tries to make some efforts for it.
Generally, in continuous-time stochastic jumping systems, the sojourn time is the
time duration between the two jumps. The sojourn time h is a random variable fol-
lowing continuous probability distribution F, for instance; F is an exponential dis-
tribution in the MJSs. Correspondingly, the transition rate λij (h), which is largely
determined by F, refers to the speed that the system jumps from mode i to mode j.
For example, if F is an exponential distribution, then λij (h) ≡ λij is a constant, which
derives from the memoryless property of the exponential distribution. As pointed in
Ref. [62], only the exponential distribution among the continuous-time probability
distributions pertains the memoryless property, which indicates that the jump speed
of the stochastic process is independent of the past. Therefore, using MJSs to model
stochastic system requires that the transition rate of the system is independent of the
past. Due to the exponential distribution for the sojourn time, the transition rate from
mode i to mode j is constant, that is,

fij (t ) λije − λijt

λij = = = λij ,
1 − Fij (t ) 1 − (1 − e − λijt )

in which fij (t ) is a probability distribution function from mode i to mode j, and Fij (t )
is the corresponding cumulative distribution function.
Regarding the semi-Markovian process, for example, the probability distribution
of sojourn time follows Weibull distribution: for a given shape parameter α > 0 and a
scale parameter β > 0, the probability distribution function is in the form

 α α −1 h
 α h exp[−( )α ], h ≥ 0,
f (h) =  β β
 0, h < 0,


The cumulative distribution function is known as

 h α
 1 − exp[−( ) ], h ≥ 0,
F (h) =  β
 0, h < 0,

10 Sliding Mode Control of Semi-Markovian Jump Systems

Therefore, the transition rate function λ (h) can be computed as

f (h) α α −1
λ (h) = = h .
1 − F (h) β α

As a result of the relaxation, the memoryless property in the MJSs does not pertain in
the S-MJSs, so the transition rate is not a constant in the S-MJS. This poses the main
technical difficulty for the stochastic stability analysis for the S-MJSs. Also, we can
see that MJSs are a special kind of S-MJSs with α = 1.
On the complete probability space (Ω, F , P ), consider the following linear S-MJS:

 x(t ) = A(rt ) x (t ) + B(rt )u(t )

 . (1.4)
 x (t ) = φ (t ),

where x (t ) is the system-state vector, and A(rt ) are the system matrices with
­compatible dimensions that depend on rt . {rt , t ≥ 0} is a continuous-time homogeneous
semi-Markovian process taking values in a finite space  = {1,2,, s}.

Definition 1.4 [63]

The evolution of the semi-Markovian process {rt , t ≥ 0} is governed by the following

probability transitions:

 π ij (h)h + o(h), i ≠ j,
Pr{rt + h = j | rt = i} = 
1 + π ii (h)h + o(h), i = j,


where h > 0 and lim o(h) / h = 0 , π ij (h) > 0, m ≠ n , is the transition rate from mode m
h→ 0

at time t to mode n at time t + h, and π ii (h) = − ∑π (h) < 0 for all i ∈ .

j ≠i
ij

In addition, for the system (1.4), we have the following lemma.

Lemma 1.2 [63]

The linear S-MJS (1.4) is stochastically stable if there exists positive-definite matrix
Pi such that the following condition is satisfied
s

Pi Ai + AiT Pi + ∑π (h)P < 0.

j =1
ij j

Based on the above lemma, a lot of research results have been proposed. The literature
[64,65] proposed the stability analysis for a class of phase-type S-MJS, in which the
sojourn time obeyed the phase-type distribution. The stochastic stability and robust-
state feedback stabilization of S-MJSs were discussed by LMI method in Ref. [66].
Introduction 11

In Ref. [63], the time-varying transition rates of the system were divided into M equal
parts by using the segmentation method, and a new criterion for the stochastic stabil-
ity of the system was given, which greatly reduced the conservatism of the conclusion.
In Refs. [67,68], the authors studied the generalized moment stability and stochastic
stabilization of linear S-MJSs, respectively. In Ref. [69], Li and Shi systematically
proposed the stochastic stability analysis, the estimated state-based SMC design, out-
put feedback quantitative control analysis, filtering design and fault diagnosis of the
S-MJSs. Zhang [70] studied the stochastic stability and stabilization of linear discrete-
time S-MJSs by using the semi-Markovian kernel method. Through the matrix spec-
tral radius analysis, the stochastic stability analysis of the positive S-MJSs was given
in Ref. [71]. In Ref. [72], the SMC approach was applied to study the passive control
of S-MJSs in the presence of actuator failure, in which the research was generally
based on uncertain transition rate. The application of the semi-Markovian model in
cluster synchronous traction control of complex dynamic networks was investigated
in Ref. [73], etc. Many scholars have made their contributions to the research of SMC
for S-MJSs: for example, as discussed in Refs. [74–79], the design of sliding mode
observer, adaptive law and adaptive SMC strategy with switching rules were proposed
for S-MJSs or S-MJSs with time-varying delay. For other researches of S-MJSs, such
as neural network-based filtering design, stochastic synchronization analysis, fault-
tolerant control and finite-time control, please refer to [80–87] and references therein.
For S-MJSs, it is difficult to analyze the performance indexes of the system because
of time-varying transition rates, which may show high nonlinearity. Some scholars
have made in-depth research on this issue. For example, in Ref. [63], the authors set
the value of time-varying transition rates upper- and lower-bounded, and then used the
known upper and lower bounds to give the stochastic stability criterion, respectively.
Following the method, in Ref. [88], the authors linearized the time-varying transition
rates by linear combination functions. In Ref. [83], the authors assumed that the dwell
time of the system modes obeyed the completely known probability distributions,
such as the Gaussian distribution and Weibull distribution, etc., and thus obtained the
known transition rate matrix, based on which the criterion conditions for checking
stochastic stability were obtained. More generally, in Ref. [89], the authors for the first
time gave the stochastic admissibility of generalized S-MJSs with uncertain transi-
tion rates and partially unknown transition rates, which laid a solid foundation for
subsequent theoretical research and extension of S-MJSs. However, the studies in Ref.
[89] still have some deficiencies: for example, how to deal with the stability and stabi-
lization issues when the jumping information from one mode to others is completely
unknown. Therefore, the study of the S-MJSs should be further discussed.

1.3 
PREVIEW OF THIS BOOK
The contents of this book are organized as follows:
Chapter 1 gives the research background, motivations and research problems
of this book, which mainly involve SMC methodologies in continuous-time and
­discrete-time, application and modeling of MJSs and S-MJSs. A survey is provided
on the fundamental theory of the SMC methodologies, which include some basic
concepts, sliding surface design, sliding mode controller design and chattering
12 Sliding Mode Control of Semi-Markovian Jump Systems

problems. Then, an overview of recent developments of S-MJSs is presented, which

includes comparisons with MJSs and latest results on S-MJSs. Finally, we summa-
rize the main contributions of this book and give the outline of this book.
Chapter 2 is devoted to provide further criterion for stochastic stability analysis
of semi-Markovian jump linear systems (S-MJLSs), in which more generic transition
rates will be studied. As is known, the time-varying TR is one of the key issues to be
considered in the analysis of S-MJLS. Therefore, this article is to investigate general
cases for the TRs that covered almost all types, especially for the type that the jumping
information from one mode to another is fully unknown, which is merely investigated
before. By virtue of stochastic functional theory, sufficient conditions are developed
to check stochastic stability of the underlying systems via linear matrix inequalities
formulation combined with a maximum optimization algorithm. Finally, numerical
examples are given to verify the validity and effectiveness of the obtained results.
Chapter 3 addresses the issue of robust fuzzy SMC for continuous-time nonlin-
ear Takagi–Sugeno fuzzy systems with semi-Markovian switching. The focus is
on designing a novel fuzzy integral sliding surface without assuming that the input
matrices are the same with full column rank and then developing a fuzzy sliding
mode controller for stochastic stability purpose. Based on Lyapunov theory, a set of
newly developed LMI conditions are established for stochastic stability of the sliding
mode dynamics with generally uncertain transition rates, and then extended to where
the input matrix is plant-rule-independent, as discussed in most existing literature.
Furthermore, finite-time reachability of the sliding surface is also guaranteed by the
proposed fuzzy sliding mode control laws. A practical example is provided to dem-
onstrate the effectiveness of the established method numerically.
Chapter 4 is concerned with finite-time SMC of continuous-time S-MJSs with
immeasurable premise variables via fuzzy approach. First, an integral sliding sur-
face is constructed based on fuzzy observer. Second, an observer-based SMC law
is synthesized to guarantee finite-time reachability of the predefined sliding sur-
face before the prescribed time. Third, through finite-time boundedness analysis, the
required boundedness performance is conducted at the reaching phase first and then
the sliding motion phase, respectively. Furthermore, sufficient conditions in terms
of linear matrix inequalities are established to guarantee the required boundedness
performance of the overall closed-loop controlled system during the two phases with
generally uncertain transition rates simultaneously. Finally, a practical example is
given to show the validity of the established method numerically.
Chapter 5 deals with the issue of observer-based adaptive SMC of nonlinear
Takagi–Sugeno fuzzy systems with semi-Markovian switching and immeasurable
premise variables. More general nonlinear systems are described in the model since
the selections of premise variables are the states of the system. First, a novel integral
sliding surface function is proposed on the observer space; then the sliding mode
dynamics and error dynamics are obtained in accordance with estimated premise
variables. Second, sufficient conditions for stochastic stability with an H ∞ perfor-
mance disturbance attenuation level γ of the sliding mode dynamics with different
input matrices are obtained based on generally uncertain transition rates. Third, an
observer-based adaptive controller is synthesized to ensure the finite-time reachabil-
ity of a predefined sliding surface. Finally, the single-link robot arm model is pro-
vided to verify the control scheme numerically.
Introduction 13

Chapter 6 proposes a decentralized adaptive SMC scheme for the stabilization of

large-scale semi-Markovian jump-interconnected systems, in which dead-zone lin-
earity in the input and unknown interconnections among subsystems are to be dealt
with. By designing integral sliding surface for each subsystem, local sliding mode
dynamics is obtained, which has good property of dynamics. Based on the LMI tech-
nique, sufficient conditions are established for checking the stochastic stability of the
sliding mode dynamics under generally uncertain transition rates. The developed
local adaptive sliding mode laws not only guarantee finite-time reachability of slid-
ing surface, but also compensate the effects from dead-zone nonlinearity in the input
and unknown interconnections among subsystems. Finally, a numerical example is
proposed to verify the effectiveness of the control scheme.
Chapter 7 concerns the problem of SMC design for nonlinear switching semi-
Markovian jump delayed systems. By choosing a linear switching surface function,
we first derive reduced-order sliding mode dynamics. Then, the property of sliding
mode dynamics with generally uncertain transition rates is analyzed by checking
a set of new conditions. Further, an adaptive SMC law is constructed to ensure the
finite-time reaching condition. Finally, practical examples are provided to dissemi-
nate the effectiveness of the proposed method numerically.

1.4 
SOME USEFUL DEFINITIONS AND LEMMAS
Lemma 1.3 [90]

(Schur complement) Consider a symmetric matrix Q such that

 Q11 Q12 
Q= 
 Q21 Q22 

(i) Q > 0 if and only if

 Q11 > 0,
 T −1
Q − Q12 Q11 Q12 > 0
 22

Or

 Q22 > 0,

Q − Q12Q11−1Q12
T
>0
 11

Lemma 1.4 [91]

Given any real number ε and any square matrix R, the matrix inequality

ε ( R + RT ) ≤ ε 2 F + RF −1RT

holds for any matrix F > 0.

14 Sliding Mode Control of Semi-Markovian Jump Systems

Lemma 1.5 [92]

For any vector x , y ∈ n, 0 < P ∈  n× n , D ∈  n× n f , E ∈  n× n f and F (t ) ∈  n f × n f satisfies

F T (t ) F (t ) ≤ I ; then for any scalar ε > 0 and matrix Q, the following inequalities hold:

1. 2 x T y ≤ x T Px + yT P −1 y,
2. Q + DF (t ) E + E T F T (t ) D T ≤ Q + ε −1DD T + ε E T E.

Definition 1.5 [93]

Given the Lyapunov functional candidate V ( x (t ), rt , t ≥ 0), which is twice differen-

tiable on x (t ). Then, its infinitesimal operator V ( x (t ), rt ) is defined by

E{V ( x (t + δ ), rt +δ ) | x (t ), rt = m)} − V ( x (t ), m)
V ( x (t ), rt ) = lim .
δ →0 δ

1.5 
ABBREVIATIONS AND NOTATIONS
 integer
 n or  n× m field of n-dimensional real vector or n × m real matrices
Ω sample space
 σ -algebra
 probability measurement
(Ω, F , P ) complete probability space
⋅ expectation operator with respect to probability measures
E{}
(⋅) infinitesimal operator of the Lyapunov function
L 2 [0, +∞) space of square integrable functions on [0, +∞)
P > 0 ( P ≥ 0) P is a symmetric positive (semi-positive) definite matrix
P − Q is a symmetric positive-definite matrix
AT or A−1 transpose of matrix A or inverse of matrix A
I or 0 identity matrix or zero matrix with appropriate dimension
∅ empty set
| ⋅ | Euclidean vector norm
 ⋅  Euclidean matrix norm (spectral norm)
 ⋅ 2 L 2-norm

∑ sum
∀ for all
Introduction 15

 is defined as
∗ symmetric terms of a matrix
Tr( A) trace of the square matrix A
rank( A) rank of matrix A
diag{a1 , a2 ,…, an} diagonal matrix with diagonal elements a1 , a2 ,, an
⋅ sign function
sgn{}
det(⋅) matrix determinant
min(⋅) or max(⋅) takes the maximum or minimum value
λmin (⋅) or λmax (⋅) minimum or maximum eigenvalue of a matrix
He( P) defined as P + P
T

lim limit
sup supremum
inf infimum

REFERENCES
[1] S. V. Emelyanov, Variable Structure Control Systems, Moscow: Nauka (in Russia),
1970.
[2] Y. Itkis, Control Systems of Variable Structure. New York: Wiley, 1976.
[3] A. Khalid, J. X. Xu, and X. Yu, On the discrete-time integral sliding-mode control,
IEEE Transactions on Automatic Control, Vol. 52, No. 4, pp. 709–715, 2007.
[4] S. V. Drakunov, V. I. Utkin, On discrete-time sliding modes, Preprints of the IFAC
Symposium on Nonlinear Control System Design, 1989. Pergamon, 1990, pp. 273–278.
[5] J. A. Burton and A. S. I. Zinober, Continuous approximation of variable structure con-
trol, International Journal of Systems Science, Vol. 17, No. 6, pp. 875–885, 1986.
[6] S. C. Y. Chung and C. L. Lin, A transformed Lure problem for sliding mode control
and chattering reduction, IEEE Transactions on Automatic Control, Vol. 44, No. 3,
pp. 563–568, 1999.
[7] V. Parra-Vega and G. Hirzinger, Chattering-free sliding mode control for a class of non-
linear mechanical systems, International Journal of Robust and Nonlinear Control:
IFAC – Affiliated Journal, Vol. 11, No. 12, pp. 1161–1178, 2001.
[8] W. Gao, Theory and Design Method of Variable Structure Control, Beijing: Science and
Technology Press, 1996.
[9] J. Xu, Y. Pan and T. Lee, A gain scheduled sliding mode control scheme using filtering
techniques with applications to multilink robotic manipulators, Journal of Dynamic
Systems, Measurement, and Control , Vol. 122, No. 4, pp. 641–649, 2000.
[10] Y. Kim, Y. Han and W. You, Disturbance observer with binary control theory, PESC
Record. 27th Annual IEEE Power Electronics Specialists Conference. IEEE, Vol. 2,
pp. 1229–1234, 1996.
[11] Y. Eun, J. Kim, K. Kim, et al., Discrete-time variable structure controller with a decou-
pled disturbance compensator and its application to a CNC servomechanism, IEEE
Transactions on Control Systems Technology, Vol. 7, No. 4, pp. 414–423, 1999.
[12] A. J. Koshkouei, K. J. Burnham and A. S. I. Zinober, Dynamic sliding mode control design,
IEE Proceedings-Control Theory and Applications, Vol. 152, No. 4, pp. 392–396, 2005.
16 Sliding Mode Control of Semi-Markovian Jump Systems

[13] M. Zribi, H. Sira-Ramirez and A. Ngai, Static and dynamic sliding mode control
schemes for a permanent magnet stepper motor, International Journal of Control,
Vol. 74, No. 2, pp. 103–117, 2001.
[14] J. Lo and Y. Kuo, Decoupled fuzzy sliding-mode control, IEEE Transactions on Fuzzy
Systems, Vol. 6, No. 3, pp. 426–435, 1998.
[15] X. Yu, Z. Man and B. Wu, Design of fuzzy sliding-mode control systems, Fuzzy Sets
and Systems, Vol. 95, No. 3, pp. 295–306, 1998.
[16] K. Jezernik, M. Rodič and B. Curk, Neural network sliding mode robot control,
Robotica, Vol. 15, No. 1, pp. 23–30, 1997.
[17] Y. Li, K. C. Ng, D. J. Murray-Smith, et al. Genetic algorithm automated approach to
the design of sliding mode control systems, International Journal of Control, Vol. 63,
No. 4, pp. 721–739, 1996.
[18] M. J. Mahmoodabadi, M. Taherkhorsandi, M. Talebipour, et al. Adaptive robust PID
control subject to supervisory decoupled sliding mode control based upon genetic algo-
rithm optimization, Transactions of the Institute of Measurement and Control, Vol. 37,
No. 4, pp. 505–514, 2015.
[19] J. Yang, S. Li and X. Yu, Sliding-mode control for systems with mismatched uncertain-
ties via a disturbance observer, IEEE Transactions on Industrial Electronics, Vol. 60,
No. 1, pp. 160–169, 2012.
[20] H. H. Choi, An explicit formula of linear sliding surfaces for a class of uncertain dynamic
systems with mismatched uncertainties, Automatica, Vol. 34, No. 8, pp. 1015–1020, 1998.
[21] F. Harashima, H. Hashimoto and K. Maruyama, Sliding mode control of manipulator
with time-varying switching surfaces, Transactions of the Society of Instrument and
Control Engineers, Vol. 22, No. 3, pp. 335–342, 1986.
[22] Y. Niu, D. W. C. Ho and J. Lam, Robust integral sliding mode control for uncertain
stochastic systems with time-varying delay, Automatica, Vol. 41, No. 5, pp. 873–880,
2005.
[23] Y. Feng, X. Yu and Z. Man, Non-singular terminal sliding mode control of rigid manip-
ulators, Automatica, Vol. 38, No. 12, pp. 2159–2167, 2002.
[24] H. Pang and G. Tang, Global robust optimal sliding mode control for a class of uncer-
tain linear systems, 2008 Chinese Control and Decision Conference. IEEE, pp. 3509–
3512, 2008.
[25] K. Furuta, Sliding mode control of a discrete system, Systems & Control Letters,
Vol. 14, No. 2, pp. 145–152, 1990.
[26] X. Su, X. Liu, P. Shi, et al. Sliding mode control of discrete-time switched systems
with repeated scalar nonlinearities, IEEE Transactions on Automatic Control, Vol. 62,
No. 9, pp. 4604–4610, 2016.
[27] D. Ginoya, P. Shendge and S. Phadke, Sliding mode control for mismatched uncer-
tain systems using an extended disturbance observer, IEEE Transactions on Industrial
Electronics, Vol. 61, No. 4, pp. 1983–1992, 2013.
[28] C. Wen and C. Cheng, Design of sliding surface for mismatched uncertain systems
to achieve asymptotical stability, Journal of the Franklin Institute, Vol. 345, No. 8,
pp. 926–941, 2008.
[29] L. Wu, X. Su and P. Shi, Sliding mode control with bounded L2 gain performance of
Markovian jump singular time-delay systems, Automatica, Vol. 48, No. 8, pp. 1929–
1933, 2012.
[30] X. Zhao, H. Yang, W. Xia, et al. Adaptive fuzzy hierarchical sliding-mode control for a
class of MIMO nonlinear time-delay systems with input saturation, IEEE Transactions
on Fuzzy Systems, Vol. 25, No. 5, pp. 1062–1077, 2016.
[31] Y. Feng, X. Yu and F. Han. On nonsingular terminal sliding-mode control of nonlinear
systems, Automatica, Vol. 49, No. 6, pp. 1715–1722, 2013.
Introduction 17

[32] J. Yang, J. Wu and A. K. Agrawal, Sliding mode control for nonlinear and hysteretic
structures, Journal of Engineering Mechanics, Vol. 121, No. 12, pp. 1330–1339, 1995.
[33] L. Wu, Y. Gao, J. Liu, et al. Event-triggered sliding mode control of stochastic systems
via output feedback, Automatica, Vol. 82, pp. 79–92, 2017.
[34] M. Liu, L. Zhang, P. Shi P, et al., Robust control of stochastic systems against bounded
disturbances with application to flight control, IEEE transactions on Industrial
Electronics, Vol. 61, No. 3, pp. 1504–1515, 2013.
[35] F. Plestan, Y. Shtessel, V. Bregeault, et al., New methodologies for adaptive sliding
mode control, International Journal of Control, Vol. 83, No. 9, pp. 1907–1919, 2010.
[36] G. Bartolini, A. Ferrara and V. I. Utkin, Adaptive sliding mode control in discrete-time
systems, Automatica, Vol. 31, No. 5, pp. 769–773, 1995.
[37] H. Li, P. Shi, D. Yao, et al. Observer-based adaptive sliding mode control for nonlinear
Markovian jump systems, Automatica, Vol. 64, pp. 133–142, 2016.
[38] J. Liu, S. Vazquez, L. Wu, et al. Extended state observer-based sliding-mode control for
three-phase power converters, IEEE Transactions on Industrial Electronics, Vol. 64,
No. 1, pp. 22–31, 2016.
[39] Y. Guo and P. Y. Woo, An adaptive fuzzy sliding mode controller for robotic manip-
ulators, IEEE Transactions on Systems, Man, and Cybernetics-Part A: Systems and
Humans, Vol. 33, No. 2, pp. 149–159, 2003.
[40] M. Y. Hsiao, T. H. S. Li, J. Z. Lee, et al. Design of interval type-2 fuzzy sliding-mode
controller, Information Sciences, Vol. 178, No. 6, pp. 1696–1716, 2008.
[41] B. S. Park, S. J. Yoo, J B. Park, et al. Adaptive neural sliding mode control of nonholo-
nomic wheeled mobile robots with model uncertainty, IEEE Transactions on Control
Systems Technology, Vol. 17, No. 1, pp. 207–214, 2008.
[42] P. Lin, C. F. Hsu, T. T. Lee, et al. Robust fuzzy-neural sliding-mode controller design
via network structure adaptation, IET Control Theory & Applications, Vol. 2, No. 12,
pp. 1054–1065, 2008.
[43] B. Beltran, T. Ahmed-Ali and M. Benbouzid, High-order sliding-mode control of
­variable-speed wind turbines, IEEE Transactions on Industrial Electronics, Vol. 56,
No. 9, pp. 3314–3321, 2008.
[44] V. I. Utkin, Discussion aspects of high-order sliding mode control, IEEE Transactions
on Automatic Control, Vol. 61, No. 3, pp. 829–833, 2015.
[45] Y. Xia, Z. Zhu and M. Fu. Back-stepping sliding mode control for missile systems
based on an extended state observer, IET Control Theory & Applications, Vol. 5, No. 1,
pp. 93–102, 2011.
[46] S. Dadras, S. Dadras and H. R. Momeni, Linear matrix inequality based fractional
integral sliding-mode control of uncertain fractional-order nonlinear systems, Journal
of Dynamic Systems, Measurement, and Control, Vol. 139, No. 11, pp. 1–7 2017.
[47] J. Liu and X. Wang, Advanced Sliding Mode Control for Mechanical Systems, Beijing:
Springer, 2012.
[48] V. Ugrinovskii and H. R. Pota, Decentralized control of power systems via robust con-
trol of uncertain Markov jump parameter systems, International Journal of Control,
Vol. 78, No. 9, pp. 662–677, 2005.
[49] C. Andrieu, M. Davy and A. Doucet, Efficient particle filtering for jump Markov
systems. Application to time-varying autoregressions, IEEE Transactions on Signal
Processing, 2003, Vol. 51, No. 7, pp. 1762–1770, 2003.
[50] J. Bai and P. Wang, Conditional Markov chain and its application in economic time
series analysis, Journal of Applied Econometrics, Vol. 26, No. 5, pp. 715–734, 2011.
[51] R. S. Ellis, Large deviations for the empirical measure of a Markov chain with an appli-
cation to the multivariate empirical measure, The Annals of Probability, Vol. 16, No. 4,
pp. 1496–1508, 1988.
18 Sliding Mode Control of Semi-Markovian Jump Systems

[52] N. N. Krasovskii, and E. A. Lidskii, Analysis design of controllers in systems with

random attributes. Part I. Automation and Remote Control, Vol. 22, pp. 1021–1025,
1961.
[53] J. G. Kemeny and J. L. Snell, Markov Chains, New York: Springer-Verlag, 1976.
[54] O. L. D. Valle Costa, M. D. Fragoso and M. G. Todorov, Continuous-time Markov Jump
Linear Systems, Berlin Heidelberg: Springer Science & Business Media, 2012.
[55] L. Hu, P. Shi and P. M. Frank, Robust sampled-data control for Markovian jump linear
system Berlin Heidelberg s, Automatica, Vol. 42, No. 11, pp. 2025–2030, 2006.
[56] F. Kozin, On relations between moment properties and almost sure Lyapunov stabil-
ity for linear stochastic systems, Journal of Mathematical Analysis and Applications,
Vol. 10, No. 11, pp. 324–353, 1965.
[57] E. K. Boukas, Stochastic Switching Systems: Analysis and Design, Boston, MA:
Birkhäuser, 2007.
[58] J. Janssen and R. Manca, Applied Semi-Markov Processes, Boston, MA: Springer
Science & Business Media, 2006.
[59] V. S. Barbu and N. Limnios, Semi-Markov Chains and Hidden Semi-Markov Models
Toward Applications:Their Use in Reliability and DNA Analysis, New York: Springer
Science & Business Media, 2009.
[60] J. Wang, A. Hertzmann and D. J. Fleet, Gaussian process dynamical models, in Y.
Weiss and B. Schölkopf and J. Platt (Eds.), Advances in Neural Information Processing
Systems, pp. 1441–1448, 2006.
[61] H. Rinne, The Weibull Distribution: A Handbook, Boca Raton: Chapman and Hall/
CRC, 2008.
[62] S. M. Ross, Introduction to Probability Models. London: Academic Press, 2006.
[63] J. Huang and Y. Shi, H∞ state-feedback control for semi-Markov jump linear systems
with time-varying delays, Journal of Dynamic Systems, Measurement, and Control,
Vol. 135, No. 4, ArticleID 041012, 2013.
[64] Z. Hou, J. Luo, P. Shi P, et al. Stochastic stability of Itô differential equations with semi-
Markovian jump parameters, IEEE Transactions on Automatic Control, Vol. 51, No. 8,
pp. 1383–1387, 2006.
[65] Z. Hou, J. Luo and P. Shi, Stochastic stability of linear systems with semi-Markovian
jump parameters, The ANZIAM Journal, Vol. 46, No. 3, pp. 331–340, 2005.
[66] J. Huang and Y. Shi, Stochastic stability and robust stabilization of semi-Markov jump
linear systems, International Journal of Robust and Nonlinear Control, Vol. 23, No. 18,
pp. 2028–2043, 2013.
[67] S. H. Kim, Stochastic stability and stabilization conditions of semi-Markovian jump
systems with mode transition-dependent sojourn-time distributions, Information
Sciences, Vol. 385, pp. 314–324, 2017.
[68] H. Schioler, M. Simonsen and J. Leth, Stochastic stability of systems with semi-­
Markovian switching, Automatica, Vol. 50, No. 11, pp. 2961–2964, 2014.
[69] F. Li, P. Shi and L. Wu, Control and Filtering For Semi-Markovian Jump Systems,
Cham: Springer, 2017.
[70] L. Zhang, Y. Leng and P. Colaneri, Stability and stabilization of discrete-time semi-
Markov jump linear systems via semi-Markov kernel approach, IEEE Transactions on
Automatic Control, Vol. 61, No. 2, pp. 503–508, 2016.
[71] M. Ogura and C. F. Martin, Stability analysis of positive semi-Markovian jump linear
systems with state resets, SIAM Journal on Control and Optimization, Vol. 52, No. 3,
pp. 1809–1831, 2014.
[72] B. Jiang, Y. Kao, C. Gao, et al. Passification of uncertain singular semi-Markovian
jump systems with actuator failures via sliding mode approach, IEEE Transactions on
Automatic Control, Vol. 62, No. 8, pp. 4138–4143, 2017.
Introduction 19

[73] T. H. Lee, Q. Ma, S. Xu, et al. Pinning control for cluster synchronisation of complex
dynamical networks with semi-Markovian jump topology, International Journal of
Control, Vol. 88, No. 6, pp. 1223–1235, 2015.
[74] F. Li, L. Wu, P. Shi, et al. State estimation and sliding mode control for semi-Markovian
jump systems with mismatched uncertainties, Automatica, Vol. 51, pp. 385–393, 2015.
[75] Y. Wei, J. H. Park, Qiu J, et al. Sliding mode control for semi-Markovian jump systems
via output feedback, Automatica, Vol. 81, pp. 133–141, 2017.
[76] Q. Zhou, D. Yao, J. Wang, et al. Robust control of uncertain semi-Markovian jump
systems using sliding mode control method, Applied Mathematics and Computation,
Vol. 286, pp. 72–87, 2016.
[77] W. Qi, J. H. Park, J. Cheng, et al. Robust stabilisation for non-linear time-delay semi-
Markovian jump systems via sliding mode control, IET Control Theory & Applications,
Vol. 11, No. 10, pp. 1504–1513, 2017.
[78] W. Qi, G. Zong and H. R. Karimi. Observer-based adaptive SMC for nonlinear uncertain
singular semi-Markov jump systems with applications to DC motor, IEEE Transactions
on Circuits and Systems I: Regular Papers, Vol. 65, No. 9, pp. 2951–2960, 2018.
[79] B. Jiang, H. R. Karimi, Y. Kao, et al. Reduced-order adaptive sliding mode control
for nonlinear switching semi-Markovian jump delayed systems, Information Sciences,
Vol. 477, pp. 334–348, 2019.
[80] F. Li, P. Shi, L. Wu, et al. Quantized control design for cognitive radio networks mod-
eled as nonlinear semi-Markovian jump systems, IEEE Transactions on Industrial
Electronics, Vol. 62, No. 4, pp. 2330–2340, 2015.
[81] P. Shi, F. Li, L. Wu, et al. Neural network-based passive filtering for delayed neutral-
type semi-Markovian jump systems, IEEE Transactions on Neural Networks and
Learning Systems, Vol. 28, No. 9, pp. 2101–2114, 2017.
[82] F. Li, L. Wu and P. Shi, Stochastic stability of semi-Markovian jump systems with
mode-dependent delays, International Journal of Robust and Nonlinear Control,
Vol. 24, No. 18, pp. 3317–333, 2014.
[83] Y. Wei, J. H. Park, H.R. Karimi, et al. Improved stability and stabilization results
for stochastic synchronization of continuous-time semi-Markovian jump neural net-
works with time-varying delay, IEEE Transactions on Neural Networks and Learning
Systems, Vol. 29, No. 6, pp. 2488–2501, 2018.
[84] Y. Wei, J. Qiu, H. R. Karimi, et al. A novel memory filtering design for semi-­Markovian
jump time-delay systems, IEEE Transactions on Systems, Man, and Cybernetics:
Systems, Vol. 48, No. 12, pp. 2229–2241, 2018.
[85] L. Chen, X. Huang, and S. Fu, Observer-based sensor fault-tolerant control for semi-
Markovian jump systems, Nonlinear Analysis: Hybrid Systems, Vol. 22, pp. 161–177,
2016.
[86] K. Liang, M. Dai, H. Shen, et al. L2 − L ∞ synchronization for singularly perturbed
complex networks with semi-Markov jump topology, Applied Mathematics and
Computation, Vol. 321, pp. 450–462, 2018.
[87] X. Liu, X. Yu, X. Zhou, et al. Finite-time H∞ control for linear systems with semi-
Markovian switching, Nonlinear Dynamics, Vol. 85, No. 4, pp. 2297–2308, 2016.
[88] H. Shen, L. Su, and J. H. Park, Reliable mixed H∞ /passive control for T-S fuzzy delayed
systems based on a semi-Markov jump model approach. Fuzzy Sets and Systems,
Vol. 314, pp. 79–98, 2017.
[89] B. Jiang, Y. Kao, H. R. Karimi, et al. Stability and stabilization for singular switch-
ing semi-Markovian jump systems with generally uncertain transition rates, IEEE
Transactions on Automatic Control, Vol. 63, No. 11, pp. 3919–3926, 2018.
[90] F. Zhang, The Schur Complement and Its Applications, New York: Springer Science &
Business Media, 2006.
20 Sliding Mode Control of Semi-Markovian Jump Systems

[91] J. Xiong, and J. Lam, Robust H2 control of Markovian jump systems with uncertain
switching probabilities, International Journal of Systems Science, Vol. 40, No. 3,
pp. 255–265, 2009.
[92] Y. Wang, L. Xie, and C. E. D. Souz, Robust control of a class of uncertain nonlinear
systems, Systems & Control Letters, Vol. 19, No. 2, pp. 139–149, 1992.
[93] S. Meyn and R. Tweedie, Markov chains and stochastic stability, Springer Science &
Business Media, 2012.
 2 Stochastic Stability
of Semi-Markovian
Jump Systems with
Generally Uncertain
Transition Rates

2.1 
INTRODUCTION
It is known that transition rate is one of the most important roles that regulates
the overall dynamic performance of S-MJSs, which distinguishes the dynamics
of S-MJSs from that of deterministic switching systems. However, due to the high
complexity of real-world circumstance, obtaining the exact knowledge of transi-
tion rates seems impossible. Therefore, it is necessary to undertake an analysis of
S-MJSs with deficient-mode information. So far, it has witnessed some inspiring
works. For example, in Ref. [1], the transition rates were set to follow fully known
probability distributions, such as the Gaussian distribution. In Ref. [2], the authors
confined the transition rates π ij (h) lower- and upper-bounded. And in Ref. [3], a
phase-type semi-Markov process was transformed into its associated Markov
chain. However, it is noted that all the results proposed in the above literature
required information on all transition rates, either full or partial cases. Therefore,
the analysis and synthesis of S-MJSs with generally uncertain TRs becomes a hot
issue. It has witnessed some results in the field of MJSs: for example, the stability
and stabilization problems for a class of continuous-time and discrete-time linear
MJS with partly unknown transition probabilities were investigated in Ref. [4],
which covers completely known and completely unknown transition probabilities
as two special cases. Also in Ref. [5], the stability and stabilization problems for
a kind of continuous-time and discrete-time MJS with generally bounded transi-
tion rates (probability) were studied. Recently, some new results are proposed
in Ref. [6] with a novel method to deal with generally uncertain time-varying
transition rates, that is, some transition rates are fully unknown. But in Ref. [6],
only two cases for the mode information in the transition rate matrix are inves-
tigated, which is far from enough since some important and difficult issues have
not been touched yet, especially for the types that transition rates from one mode
to others are totally unknown. Although some methods have been proposed in

21
22 Sliding Mode Control of Semi-Markovian Jump Systems

dealing with MJSs with deficiency-mode information [7], the techniques cannot
be extended to S-MJSs directly or might be invalid for S-MJSs. So far, to the best
of our knowledge, the issue of stochastic stability analysis for S-MJSs with more
generic transition rates is an open challenge that has not been addressed, which
motivates us to do the research.
Based on the above analysis, this chapter deals with the problem of stochastic
stability analysis for linear S-MJSs with generic transition rates, which includes full
information, partial information and totally unknown information about the modes.
The main contribution is that a set of feasible criteria are established to check the
stochastic stability of the considered systems. Particularly, for the case that all transi-
tion rates from one mode to others are fully unknown, a feasible estimation method is
proposed to deal with this issue. Then, by solving a maximum optimization problem,
feasible solutions for the established linear matrix inequality (LMI) conditions can
be obtained in the sense of stochastic stability.

2.2 
SYSTEM DESCRIPTION
Consider the following continuous-time linear S-MJS fixed on the probability space
(Ω, F , P ):

x (t ) = A(rt ) x (t ),
(2.1)
x (0) = x 0 , r (0) = r0 ,

where x (t ) ∈  n is the state vector, A(rt ) is the mode-dependent system matrix with
appropriate dimensions, x 0 is the initial condition, and r0 is the initial mode of the
semi-Markov process. {rt , t ≥ 0} is a semi-Markov process taking values in a finite set
 = {1,2,…, s} and governed by

 π ij (h)h + o(h), i ≠ j,
Pr{rt + h = j | rt = i} =  (2.2)
1 + π ii (h)h + o(h), i = j,


where h > 0 and lim o(h) / h = 0 , π ij (h) > 0, i ≠ j , is the transition rate from mode i at
h→ 0

time t to mode j at time t + h, and π ii (h) = − ∑ π (h) < 0 for each i ∈. In addition,
j ≠i
ij

A(rt ) will be denoted by Ai in the following.

Based on the above discussion, the transition rate matrix for the system (2.1) can
be described as

 π 11 (h) π 12 (h)  π 1s (h) 

 
π 21 (h) π 22 (h)  π 2 s (h)
Π= . (2.3)
     
 
π
 s1 (h) π s 2 (h)  π ss (h) 
Stochastic Stability 23

Definition 2.1 [2]

The linear S-MJS (2.1) with all modes and all t ≥ 0 is said to be stochastically stable
if there exists a finite positive constant T ( x 0 , r0 ) such that the following inequality
holds for any initial condition ( x 0 , r0 ):
t
E
∫  xˆ(s) 
0
2
ds ≤ µ −1EV ( xˆ (0), r0 ).

Remark 2.1

Generally, in order to deal with the analysis and synthesis of continuous-time S-MJSs,
the following methods have been proposed to tackle the transition rate π ij (h):

a. Huang and Shi have proposed some pioneer works [2], in which π ij (h) was
set lower- and upper-bounded as π ij (h) ∈[π ij , π ij ] with π ij and π ij being the
known lower and upper bounds of the transition rate π ij (h), respectively.
Further, in order to reduce conservativeness of the proposed stability crite-
rion, the transition rate was further separated into M sections with division
points π ij ,0 , π ij ,1 ,, π ij , M −1 , π ij , M ;
b. Based on the general assumption from Ref. [2] that π ij (h) is lower- and
upper-bounded, the following conditions are further developed in Ref. [8]:
K K

π ij (h) = ∑ζ π , ∑ζ
k =1
k ijk
k =1
k = 1,ζ k ≥ 0, (2.4)

with

 π ij − π ij
 π ij + ( k − 1) , i ≠ j, j ∈S ,
 K −1
π ijk =
 π ij − ( k − 1) π ij − π ij , i = j, j ∈S ,
 K −1

c. Particularly, the probability distribution function (PDF) of the sojourn time

h staying at one mode was selected to obey certain distribution in Ref. [1],
such as the Weibull distribution or Gaussian distribution. Then, it was com-
∞
puted that E{π ij (h)} =
∫ 0
π ij (h) gi (h)dh, where gi (h) is the PDF. Therefore,
the mathematical expectation of the transition rate matrix can be defined by

 E{π 11 (h)} E{π 12 (h)}  E{π 1s (h)} 

 
 E{π 21 (h)} E{π 22 (h)}  E{π 2 s (h)} 
(2.5)
     
 
 E{π s1 (h)} E{π s 2 (h)}  E{π ss (h)} 

24 Sliding Mode Control of Semi-Markovian Jump Systems

d. In view of the above three cases, which all need partial or full information
on the TR π ij (h), the following TR matrix in Ref. [9] was introduced to bet-
ter accommodate practical dynamical systems

 π 11 + ∆π 11 (h) ?  π 1s + ∆π 1s (h) 
 
 ? ?  π 2 s + ∆π 2 s (h) . (2.6)
     
 
 ? π s 2 + ∆π s 2 (h)  π ss + ∆π ss (h) 

This chapter will follow the case (d) and extend the result to more generic cases.
Now, revisit the case (d) in Ref. [9]. Practically, for the high complexity of the real-
world dynamics, the TRs for S-MJSs are considered to satisfy the following two
cases: (I): π ij (h) is fully unknown and (II): π ij (h) is not exactly known but upper-
and lower-bounded. Following [1] that π ij (h) ∈[π ij , π ij ], in which π ij and π ij are the
known real constants representing the lower and upper bounds of π ij (h), respec-
1
tively. Denote π ij (h)  π ij + ∆π ij (h), in which π ij = (π ij + π ij ) and | ∆π ij (h) |≤ δ ij with
2
1
δ ij = (π ij − π ij ). The TR matrix with three jumping modes may be described as
2

 π 11 + ∆π 11 (h) ? π 13 + ∆π 13 (h) 
 
 ? ? π 23 + ∆π 23 (h) , (2.7)
 ? π 32 + ∆π 32 (h) ? 
 

where “?” is the description of unknown TRs. For brevity, ∀i ∈ , let I i = I i ,kn ∪ I i ,ukn,
where I i ,kn and I i ,ukn are defined as follows:

I i , kn  { j : π ij can be determined for j ∈},

I i ,ukn  { j : π ij is not known for j ∈}.

Basically, the results developed in Ref. [9] have some limitations due to the assump-
tions that both I i ,kn and I i ,ukn are not empty. Therefore, only two cases are studied in
Ref. [9], that is, i ∈ I i ,kn with I i ,ukn ≠ ∅ and i ∈ I i ,ukn with I i ,kn ≠ ∅. Here, the most impor-
tant case to be studied is that what if i ∈ I i ,ukn while I i ,kn is empty, that is, I i ,kn = ∅. For
instance, the TR matrix with four modes is defined by the following form:

 π 11 + ∆π 11 (h) π 12 + ∆π 12 (h) π 13 + ∆π 13 (h) π 14 + ∆π 14 (h) 

 
 π 21 + ∆π 11 (h) ? π 23 + ∆π 23 (h) ?  . (2.8)
 ? π 32 + ∆π 32 (h) π 33 + ∆π 33 (h) ? 
 
 ? ? ? ? 
Stochastic Stability 25

Therefore, based on the above TR matrix, the following cases are needed to be
studied:

i. i ∈ I i ,kn and i ∈ I i ,kn is fully known, that is, I i ,kn = S ;

ii. i ∈ I i ,kn and i ∈ I i ,kn is partially known, that is, I i ,kn ≠ S while I i ,kn is also not
empty;
iii. i ∈ I i ,ukn and i ∈ I i ,kn is partially known, that is, I i ,kn ≠ S while I i ,kn is also not
empty;
iv. i ∈ I i ,ukn and i ∈ I i ,kn is fully unknown, that is, I i ,kn = ∅.

Here, we need to point out that the case (iv) will not exist for all i ∈ S in the following
investigation.
Compared with previous literature, more universal TRs (2.8) will be investigated
in this chapter. Especially for the case (iv), which is the most challenging part to
be analyzed in the sequel. Therefore, in order to give stochastic stability of semi-
Markovian jump linear system (S-MJLS) (2.1) in the sense of case (iv), the following
estimation method is proposed:
(iv-i): i ∈ I i ,ukn and I i ,kn = ∅, while there exists j ∈ I j ,kn with I j ,kn being not empty
for j ≠ i, such as the first and third rows of TR matrix (2.8). In this case, we define

π ii (h) = aiπ jj (h),

where ai is the estimated parameter to be determined.

(iv-ii): i ∈ I i ,ukn and I i ,kn = ∅, while there is no j ∈ I j ,kn for j ≠ i, that is, the cases (i)
and (ii) do not exist: for example, only the forms of the second and fourth row exist
in TR matrix (2.8). In this case, we define

π ij* (h) = aiπ jj* (h),

where π jj* (h) is the partially known j*-th TR in the j-th row of TR matrix, and ai is
the estimated parameter to be determined.
To this end, another issue comes that how to determine ai in order to investigate the
stochastic stability of an S-MJLS with TR matrix in form of (2.8). Therefore, to give a
comprehensive study of S-MJLS with generic uncertain TRs, at least five cases should
be considered, which include the cases (i)–(iii) and the case (iv) that is divided into
another two cases (iv-i) and (iv-ii). In the following, an LMI-based formulation and an
optimization algorithm are proposed to check the stochastic stability and to obtain the
estimated parameters ai . Before moving on, we denote the following nonempty set I i ,kn :

I i ,kn  { i ,1 ,  i ,2 ,…,  i ,o} 1 ≤ o < s,

where  i ,s (s ∈{1,2,…, o}) represents the index of s-th element in the i-th row of the
TR matrix.
Therefore, the overall purpose of this chapter is to recommend numerically feasible
stochastic stability conditions for the linear S-MJS (2.1) with universal generic TRs.
26 Sliding Mode Control of Semi-Markovian Jump Systems

2.3 
STOCHASTIC STABILITY ANALYSIS
Based on the above discussions, the following stochastic stability criterion is proposed
for the S-MJSs (2.1) with generic TRs in the presence of the abovementioned five cases.

Theorem 2.1

For the linear S-MJS (2.1), π ij (h) depends on sojourn time h, where h is set to 0 when
system jumps. Then the system is stochastically stable if there exist Pi > 0 and i ∈ ,
such that for all i ∈
s

AiT Pi + Pi Ai + ∑ π (h)P < 0.

j =1
ij j (2.9)

Proof: Consider the following Lyapunov functional:

V ( x (t ), r (t )) = x T (t ) P(rt ) x (t ). (2.10)

Then, according to Definition 1.5 and the method proposed in Ref. [2], we have

1
s

V ( x (t ), i) = lim 
r→0 r  ∑ Pr{ηt + r = j | ηt = i}xrT Pi xr
 j =1, j ≠ i

+ Pr{ηt + r = j | ηt = i}xrT Pi xr − x T (t ) Pi x (t )  ,


where xδ  x (t + δ ). For a general distribution of the sojourn time without memory-

less property, that is, Pr{ηt +δ = j | ηt = i} ≠ Pr{ηδ = j | η0 = i}, by the conditional prob-
ability formula, we have

1
s
qij (Gi (h + δ ) − Gi (t )) T
V ( x (t ), i) = lim 
δ →0 δ  ∑ 1 − Gi (h)
xδ Pi xδ
 j =1, j ≠ i
1 − Gi (h + δ ) T 
+ xδ Pi xδ − x T (t ) Pi x (t ) 
1 − Gi (h) 

1
s
qij (Gi (h + δ ) − Gi (h)) T
= lim 
δ →0 δ  ∑ 1 − Gi (h)
xδ Pi xδ
 j =1, j ≠ i
1 − Gi (h + δ ) T
+ [ xδ − x T (t )]Pi xδ
1 − Gi (h)
1 − Gi (h + δ ) T
+ x (t ) P T ( k )[ xδ − x (t )]
1 − Gi (h)
Gi (h + δ ) − Gi (h) T 
− x (t ) Pi x (t ) 
1 − Gi (h) 
Stochastic Stability 27

where Gi (h) is the cumulative distribution function of the sojourn time when the
system stays in mode i and qij is the probability intensity from mode i to mode j. On
the other hand, we have that

(Gi (h + δ ) − Gi (h)) 1 (G (h + δ ) − Gi (h))

lim = lim i = π i (h),
δ →0 (1 − Gi (h))δ 1 − Gi (h) δ → 0 δ

1 − Gi (h + δ )
lim = 1,
δ → 0 1 − Gi (h)

where π i (h) is the TR of the system jumping from mode i. Therefore,

s

V ( x (t ), i) = ∑ q π (h)x (t)P x(t) + 2x (t)P x(t) − π (h)x (t)P x(t).

j =1, j ≠ i
ij i
T
i
T
i i
T
i (2.11)

Now, define π ij (h) = π i (h)qij for j ≠ i and π ii (h) = − ∑ π (h). Overall, we have
j ≠i
ij

V (t ) = x T (t )Γ i x (t ). (2.12)

in which Γ i = Pi Ai + A P + T
i i ∑ π (h)P . It is seen from (2.9) that Γ
j =1
ij j i ,m < 0, Hence,

denote a scalar µ  λmin {Γ i} > 0 such that

V ( xˆ (t ), rt ) ≤ − µ  xˆ (t ) 2 . (2.13)

Therefore, by Dynkin’s formula, we get for any t > 0

t
EV ( xˆ (t ), rt ) − EV ( xˆ (0), r0 ) ≤ − µ E
∫  xˆ(s) 
0
2
ds, (2.14)

which yields
t
E
∫  xˆ(s) 
0
2
ds ≤ µ −1EV ( xˆ (0), r0 ). (2.15)

Then, it is easy to see that the linear S-MJS (2.1) is stochastically stable.
Now, based on the above results, let’s consider the more general cases.

Theorem 2.2

Given the scalar ai , the linear S-MJS (2.1) is stochastically stable if there exist
p­ ositive-definite matrices Pi > 0 , Uij > 0, Vij > 0, Wij > 0, Sij > 0 and Tij > 0 such that
the following conditions hold for each i ∈ ,
28 Sliding Mode Control of Semi-Markovian Jump Systems

Case I: i ∈ I i , kn ,I i , kn =  = {1,2,…, s},

 i11 i12 
  < 0, (2.16)
 * i13 


Case II: i ∈ I i ,kn , ∀l ∈ I i ,ukn , I i ,kn  { i ,1 ,  i ,2 ,…,  i ,o1},

 i21 i22 
  < 0, (2.17)
 * i23 


Case III: i ∈ I i ,ukn , ∀l ∈ I i ,ukn (l ≠ i), I i ,kn  { i ,1 ,  i ,2 ,…,  i ,o2 },

Pi − Pl ≥ 0, (2.18a)

 i31 i32 
  < 0, (2.18b)
 * i33 


Case IV: i ∈ I i ,ukn, there exists j ≠ i such that j ∈ I j ,kn , for ∀l ∈ I i ,ukn ,

 δ2 
 Pi Ai + AiT Pi + aiπ jj ( Pi − Pl ) + ai jj Sii ai ( Pi − Pl ) 
 4  < 0, (2.19)
 * − ai Sii 
 

Case V: i ∈ I i ,ukn, there is no j ≠ i such that j ∈ I j ,kn , for ∀l ∈ I i ,ukn (l ≠ i),

Pi − Pl ≥ 0, (2.20a)

 δ2 
 Pi Ai + AiT Pi + aiπ jj* ( Pj* − Pl ) + jj* Tij* ai ( Pj* − Pl ) 
 4  < 0, (2.20b)
 * − ai Tij* 
 
where
s s
(δ ij )2
 = Pi Ai + A P +
i
11 T
i i
j =1, j ≠ i
4∑ Uij + ∑π P ,
j =1
ij j

i12 = [( P1 − Pi ) … ( Pi −1 − Pi ) ( Pi +1 − Pi ) … ( Ps − Pi )],
i13 = [−Ui1 … −Ui (i −1) −Ui (i +1) −Uis ],

 (δ ij )2 
i21 = Pi Ai + AiT Pi + ∑  4 Vij + π ij ( Pj − Pl ) ,
j ∈Ii ,kn  
Stochastic Stability 29

i22 = [( P i ,1 − Pl ) … ( P i ,o1 − Pl )],

i23 = [−Vi i ,1 … −Vi i ,o1 ],

 (δ ij )2 
i31 = Pi Ai + AiT Pi + ∑ 
j ∈Ii ,kn
4
Wij + π ij ( Pj − Pl ) ,


i32 = [( P i ,1 − Pl ) … ( P i ,o 2 − Pl )],

i33 = [−Wi i ,1 … −Wi i ,o2 ].

Proof: Based on the former Theorem 2.1, the linear S-MJS (2.1) is stochastically
stable if it holds

AiT Pi + Pi Ai + ∑ π (h)P < 0.

j =1
ij j

However, the above inequality is not solvable in the sense of nonlinear transition
s

rates. Now, we are to deal with the term ∑ π (h)P . Let’s consider the following
j =1
ij j

cases:
Case I: i ∈ I i ,kn and I i ,kn = S.
s s

According to the partition in (2.8), let ∑

j =1, j ≠ i
π ij = −π ii . Therefore, ∑ π (h)P can
j =1
ij j

be rewritten as

s s s

∑j =1
π ij (h) Pj = ∑j =1
π ij Pj + ∑ ∆π (h)P
j =1
ij j

s s

= ∑ j =1
π ij Pj + ∑ ∆π (h)P + ∆π (h)P
j =1, j ≠ i
ij j ii i

s s

= ∑ j =1
π ij Pj + ∑ ∆π (h) ( P − P )
j =1, j ≠ i
ij j i (2.21)

s s

= ∑ j =1
π ij Pj + ∑  12 ∆π (h) ( P − P ) + 12 ∆π (h)(P − P )
j =1, j ≠ i
ij j i ij j i

s s
 δ ij2 
≤ ∑ π P + ∑  4 U
j =1
ij j
j =1, j ≠ i
ij + ( Pj − Pi )Uij−1 ( Pj − Pi ) 

30 Sliding Mode Control of Semi-Markovian Jump Systems

By applying Schur complement, it is known that (2.16) guarantees the S-MJS (2.1)
is stochastically stable in this case.
Case II: i ∈ I i ,kn, I i ,kn ≠ S and I i ,kn ≠ ∅.
First, let λi ,kn  ∑
π ij (h). Since I i ,ukn ≠ ∅, it satisfies that λi ,kn < 0 . Therefore,
j ∈Ii ,kn
s

∑ π (h)P can be written as

l =1
ij i

s  
∑π (h)P =  ∑ + ∑
j =1
ij j
j ∈Ii ,kn j ∈Ii , ukn
 π ij (h) Pj


= ∑ π (h)P − λ ∑ π−λ(h)P .
j ∈Ii ,kn
ij j i ,k
j ∈Ii ,ukn
ij

i ,k
j (2.22)

It is obvious that 0 ≤ π ij (h) / − λi ,k ≤ 1 (∀j ∈ I i ,uk ) and ∑ π−λ(h) = 1. Therefore, for

j ∈Ii ,ukn
ij

i ,k

∀l ∈ I i ,uk , it holds that

π ij (h)  
s

Pi Ai + AiT Pi + ∑ π (h) P = ∑ ij j  Pi Ai + AiT Pi +

− λi , k   ∑
π ij (h)( Pj − Pl )  . (2.23)
j =1 j ∈Ii ,ukn  j ∈Ii ,kn 
s

For 0 ≤ π ij (h) ≤ − λi ,k , Pi Ai + AiT Pi + ∑ π (h)P < 0 is equivalent to

j =1
ij j

Pi Ai + AiT Pi + ∑ π (h)(P − P ) < 0, for

j ∈Ii ,kn
ij j l ∀l ∈ I i ,ukn . (2.24)

In formula (2.24), it has

∑ π (h)(P − P ) = ∑ π (P − P ) + ∑ ∆π (h)(P − P ).
j ∈Ii ,kn
ij j l
j ∈Ii ,kn
ij j l
j ∈Ii ,kn
ij j l (2.25)

By virtue of Lemma 1.4 and for any Vij > 0 , it follows that

∑ ∆π (h)(P − P ) = ∑  12 ∆π (h)(P − P ) + 12 ∆π (h)(P − P )

j ∈Ii ,kn
ij j l
j ∈Ii ,kn
ij j l ij j l

 (δ ij )2 T
≤ ∑ −1
 4 Vij + ( Pj − Pl )(Vij ) ( Pj − Pl ) .
j ∈Ii ,kn  
(2.26)

Combining (2.22)–(2.26) and applying Schur complement, it can be deduced from

(2.17) that the S-MJS (2.1) is stochastically stable in this case.
Other documents randomly have
different content
 It is not necessary to go about with your hands
folded in an attitude of prayer, or pretend to be overly
virtuous and honest, all you need is to be a man, open
and above board, and decent in your associations with
others.
 KEEP YOUR NERVE
Most of our troubles are imaginary, nine-tenths of
them never coming to us as we expected. They are
mostly matters of nerve weakness.
We start something during the day, and lay awake
at night worrying for fear it may not turn out
successfully. We brood over phantoms and scarecrows,
for that is what most of our worries are.
If you have started anything right, and your
conscience is at rest, why do you worry? There is no
reason for it.
Or if you have used your best judgment and made
your best effort to make your venture a success, go to
rest, put your trust in God and you will sleep.
The man who loses his nerve in the middle of a
railroad or any place where there is danger, comes to
grief. The life on earth is a road full of pitfalls and
unpleasant things, many of them as dangerous as a
railroad train bearing down upon us.
If you keep your nerve, you simply get out of the
way of the locomotive, or of the runaway horse, or the
automobile, and keep on living.
So it is in your every-day transactions. Keep out of
the way of things than may undo you. Step aside and
let them pass by. Everything works for good in this
world, what you do not accomplish some other man
does, and it helps you because everything is along
parallel lines.
Where a thing is unavoidable, or inevitable, why
then it is foolish to worry, and shows poor control of
your nerves.
 Put your trust in God, follow the straight path, and
stiffen up your nerves.
 STUDY YOUR HEALTH
The ancients said that there can not be a healthy
mind in an unhealthy body. And they established this
rule for all to follow: “Keep your body healthy and your
mind will be healthy.”
By a healthy mind is meant a calm, cool, clear,
active brain that can act up to its full capacity without
faltering, or falling down at trifles.
To have that sort of brain, you must preserve your
bodily health.
One patent way to lose your bodily health is to
acquire bad habits of any kind that you know are bad
for you.
You have a headache in the morning, and no
appetite. It does not require a Solomon to tell what ails
you. You have been drinking, carousing, staying up late
instead of going to bed and getting your necessary
sleep.
You have eaten things that do not agree with you,
and so you must see a doctor. Besides that, you are too
sick to go to work.
All you have to go upon in this world are your
health and your mind. It does not matter what you do
for a living, you must keep your wits about you all the
time, and you can not do this unless you keep your
health.
The mind is so closely connected with the body
that what affects one affects the other, favorably or
unfavorably.
Eat proper food, something that you know by
experience will agree with you. Take your accustomed
sleep, and exercise your muscles to keep your nerves—
those nerves that spread up into the brain—in full play
and ready for emergencies.
A healthy man does not worry; he is an optimist
and looks at the bright side of life. An unhealthy man is
a pessimist and sees things through a dark cloud. He
ends by running down at the heels, and ceases to
possess any economical functions.
 MAKING ONE HAND WASH THE
OTHER
It is a good commercial and business maxim:
“Make one hand wash the other.”
There are little delicate attentions shown men to
induce them to do you a favor. It is not exactly doing to
others as you would have others do to you, but you do
something for a person in the expectation that he will
do something for you. This is the origin of the saying.
Politeness, forbearance and social amenities are
the rule in these days, and it is the best policy to
assume that distinction even if you do not feel that
way.
The propensity for making one hand wash the
other is more apparent in commercial and trade
transactions than in any other. It is in these
occupations that the eye beholds dollars or doughnuts
at the end of a string, and a gentle pulling in the way of
attention and brotherly reciprocation will bring the
dollars or doughnuts within reach.
Bears and dogs growl and get nasty whenever they
feel like it regardless of consequences, for they live in
the present entirely and nothing is of any importance
to them on the morrow. They do not even know enough
to lay in a supply of provisions for a rainy day. A
squirrel will do that, but squirrels are not quarrelsome,
they are friendly and gentle, they make one paw wash
the other. Watch one of them grab for a nut, get it, and
beg prettily for another.
We must provide for a rainy day, and if we are in
business we must have friends and customers to fall
back upon for shelter. Waiting until the rain sets in and
then beginning, fails—it is then too late, at least for that
day, but by beginning you will perhaps be ready for the
next rainy day.
 SUPERSTITION AND LUCK

More people are superstitious than are willing to

admit the fact. From bygone ages to modern times,
both high and low, rich and poor, educated and
ignorant, have yielded to some curious vein of fancy
that leads them to expect “luck” or success more
readily if certain whimsical conditions are complied
with. Who has not, at some time, felt the power of one
or another of the odd ideas that seem to have such a
firm hold on the mind of man? Laugh it off as we will,
declare it nonsense as we know it to be, still there is the
tendency to put an unreasoning half-belief in it.
Do we not all know those who are nervous with
fear if salt is spilled; who would go without a meal
rather than be one of thirteen at table; who never begin
any important work on a Friday; who are careful to
take their first sight of the new moon over their right
shoulder instead of the left; who rejoice in the finding
of a four-leaved clover?
“Luck” is a plant that grows from the seed. And the
seed sown is the kind of thoughts we entertain; ideas
about ourselves, about God, about our work, and about
the rest of the world.
Thoughts can be chosen. If we think ourselves
weak and inferior, we invite failure; because then the
work that we do will not be our best, and will be
surpassed in value by that of others.
 If we think instead, “I can do this work better than
it has ever been done before—and I will,” the seed will
grow and bear fruit in results to ourselves and others.
 GOOD AND BAD LUCK
Your success in life never depends upon the turn
of a card or the dice. You can neither dream yourself
into good luck, nor dream yourself out of bad luck.
Good luck keeps company only with industrious,
thrifty and honorable people who have faith in
themselves, faith in their fellow men, and faith in God.
Even then, luck will disappear like smoke in a
wind unless you can also demonstrate that you possess
wisdom, patience and courage.
What you think is good luck, may keep company
with you for a short time, but will speedily desert you if
you do not make good.
The dictionary says “Luck” means “that which
happens a person; chance; accident; good fortune;
success.”
In your luck you should keep away from the
element of “chance” or “accident.” Let your luck
depend upon your own efforts, and take things by the
forelock and make them come your way. Things will
happen you just as you intend they shall.
There is really no such thing as bad luck, for if a
thing does not happen because of your mistakes, it is
not bad luck but mistake.
Try as you may to reach a certain result, and
failing, you say you had bad luck. You merely did not
know how to succeed or went too far, or reached out
for more than you could handle. That is not bad luck, it
is mismanagement. You might have succeeded if you
had managed properly.
 Chance must be kept out of the way or you will
flounder about in a swamp whose quicksands will
engulf you sooner or later.
 BE SLOW TO ANGER
The Scripture says: “He that is slow to anger is
better than the mighty.”
We are also advised not to let the sun go down on
our wrath.
If we desire to succeed in any enterprise we must
“possess our souls in patience.” In Luke XVI, 19, it is
explained: “In your patience possess your souls.”
We are nowhere advised not to be angry, but to
possess our souls in our anger. That is: Never let anger
get the better of our control.
In Ephesians IV, 26, it is said: “Be ye angry and sin
not; let not the sun go down upon your wrath.”
This is the key to what is known as “temper.” It is
the part of a wise man to control his temper. Not to
have any temper at all is to be one whom Col.
Roosevelt calls “a mollycoddle,” and such a person is
truly weak and without any backbone.
But the anger or the temper which leads to
violence is to be controlled absolutely. Those sudden
gusts of passion lead to crime as sure as the sun rises
and sets every day. And it is always personal violence,
even to the extent of murder that is the result of giving
way to such an emotion. No one ever becomes violently
angry because he is not a good man.
If a man stands up in his manhood, and despises
small things, he will be in a position to control his
angry feelings no matter how much he may feel hurt by
the acts of another.
If we could get angry with ourselves because we do
not improve, that would be an anger worth cultivating.
But so far as others are concerned, let your anger be
mild and never reach the point of resentment, for that
always leads to revenge which is a fatal emotion.
If others are the cause of anger to us, keep away
from them, and if we must associate with them, keep
cool and bide your opportunity.
 PRACTICE LOGIC, COMMON
SENSE AND TACT
When a man can give a good reason for what he
does he practices logic. Not excuses for doing what he
should not do, but REASONS why.
When he gives good reasons, and follows the
universal practice of other men under the same
circumstances, he practices common sense.
When he does things in a quiet, unobtrusive, and
agreeable manner, so that other men are satisfied with
his way, he practices tact.
These three qualities are badges of success among
every nation and in every occupation, trade, or
profession.
The business and professional resourcefulness of
every man is not measured upon the quantity of his
learning, or his high proficiency, but according to his
ability to apply what he knows to the matter in hand.
A man may be able to measure the stars, and yet
not be able to saw a board straight. Such a man may
know much but he makes a poor carpenter.
A man should reason with himself as to the best
way of doing anything, and then do it, giving good
reasons for it.
Common sense is good judgment applied to the
every day things of life, and tact is doing those things
without disturbing others or by considering their
feelings with as much care as you do your own.
 To use a common expression: “You have got to
worm things out of the world, but you must do it as
gently as inserting a corkscrew in a stubborn cork.”
 ENCOURAGE OTHERS
When you encourage others to go ahead with what
they are doing, with a cheery word or a pleasant smile,
you are laying up treasure for yourself. For the man
you encourage will encourage you, and heaven knows
we all need encouragement.
Many men stand on the verge of a precipice of
indecision, not being able to decide whether they
should draw back or fall over.
It is not help these men want so much as it is
encouragement. They are able to help themselves but
they haven’t the nerve, and you give them a word of
cheer or encouragement, and they get right with
themselves and their work.
If a man starts into business and you can trade
with him, do so, and that will encourage him to go
ahead and strive to be successful.
He may be a beginner at manufacturing something
for the use of others. Tell him how his work or
productions are well received, or take one yourself and
use it even if you do not want it. You encourage him to
go on, and by and by you may be in a position where
you will need a little encouragement, then he will
remember you.
It is customary for the unthinking to imagine that
they must do something big or great in order to expect
returns, but this is a mistake. We show our greatness in
little things, because we know that many little things
make up a great thing. The more small things we do
the greater will be the accumulation in the end.
Do not patronize any man or he will repulse your
approach; you must encourage, which is far different
from patronizing. By assuming a patronizing air you
assume a superiority which is disliked.
This is an age of small things that go to make up
big things, and we must fall in with the conditions of
the age in which we live and expect to do business.
 HOW TO LEARN SELF-CONTROL
To master the feelings the head and the heart
should work together.
All of our emotions may be said to come from the
heart, and the latter is set in motion by the will power
which is the head.
There are times when a man feels like “boiling
over” as it is called, but policy and good judgment warn
him to keep within bounds.
It is always our sentiments or feelings and
emotions that need a curbing hand, our opinions can
take care of themselves.
Where our feelings and our mind go together there
is no trouble, for then duty and inclination go together.
But where our feelings are not regulated and
controlled, they become unstable and shifting. Like the
winds that blow where they list and whither no man
can tell, our lack of self-control may drive us to the
most violent acts. We become the sport of chance
desires and vagrant impulses.
Control is essential because from our ill-regulated
acts much injustice and harm may be done, not only to
ourselves but to others.
A man who stands above whim and caprice is a
superior in strength to a man who permits his caprices
to direct him.
What we call character has its emotions and
passions, its affections and intense sympathies, but
mastered and controlled into a whole of outward
justice and fairness.
 The true freeman fights himself free from blind
feeling and impulse; he is a happy warrior and fights
on a battlefield where his convictions and emotions are
a unit.
The Martyrs possessed such self control that
burning at the stake, or limbs torn by savage beast did
not wring a note of pain from them. “But,” you say,
“that was Divine strength.” Of course, and any one who
desires the same Divine strength to aid him control his
emotions, may have it for the asking.
 DON’T BE A DREAMER
Waste no Time Dreaming of the Past
You are living in the present preparing for the
future. The past is dead and you should let the past
bury the past.
The man who dreams of the past and forgets his
future, is like a man who rises in the morning not of
today but of yesterday. He is going backward when his
face is put in front pointing always forward.
Life is too short to be wasted in vain regrets for
what has transpired in the past. Even yesterday is
ancient history and best forgotten.
We have work to do in the present to perfect or
accomplish something in the future; it is our time of
grace, given us to grasp at opportunities as they come
before us.
While you are lamenting an opportunity that
escaped you yesterday, a better one comes along today
and passes us unnoticed.
There is too much of this sort of sorrow
experienced by the people of the earth, but when it
comes to a man with an occupation, a business man or
a young man getting ready for business, it is positively
foolish and detrimental.
We know that it has been the practice of people in
all times to fret and worry about the things of the past,
for there are numerous sayings cautioning them
against it. One of them is very appropriate: “Never cry
over spilled milk.” It is gone and can not be restored.
Many persons may have what is called a “skeleton”
in his closet, but it does not do him nor his friends any
good service to keep rattling its bones continually.
If you have been very wrong in the past, repent
and begin over again.
 DON’T BE BASHFUL
There are many persons who stand in their own
way to success by their timidity, or bashfulness.
Such people are too self-conscious, and betray
their lack of self-confidence which is regarded as an
evidence of ignorance, or at least, inability to perform
the duties they aspire to impose upon themselves.
Every man is better acquainted with himself than
anybody else, but when he relies upon the knowledge
of others as superior to his own knowledge, he loses
the respect of his fellows, and finally loses his own
respect and becomes bashful in their presence.
You should cultivate courage and exhibit
symptoms of self-confidence, for by that means you
show others that you are willing to “dare” and venture
a trial of your capacity.
If you are too timid and have no confidence in
yourself, you must not expect others to take you except
at your own valuation.
There is, however, such a thing as being over-
confident and brazen, which is the extreme of timidity,
and becomes boastfulness.
Men have a way of studying each other and
judging from their own standpoint, and if they perceive
any timidity or bashfulness, they judge against you as
incompetent. On the other hand, they quickly see
beneath the surface of boasting, and reach the same
opinion.
Be self-confident, and gentlemanly about it, for so
you will pull through any opportunity, besides making
hosts of friends in a business and social way.
 Look a man straight in the eye, but do not try to
look him down.
 DON’T BE UNDECIDED
A man who can not make up his mind to do or not
to do a thing without a great deal of wobbling first one
way and then another, is as bad as an unsafe wall in a
building—everybody keeps off lest it fall and do some
damage.
When a man has first carefully considered a
project, or a certain line of action, and also taken the
advice of his friends if the matter is important, he
should decide one way or the other at once.
A wobbly man is weak-kneed, and not to be
depended upon for any purpose.
If you have ever had dealings with that kind of a
man you will understand how painful it is to wait for
him to decide.
A man at a cross-roads hesitates and says: “Shall I
go this way or that?” He hesitates, starts, returns, starts
the other way, and finally goes the wrong way and falls
into a hole.
It has passed into a proverb that, “He who
hesitates is lost.”
Of course, there is reason and judgment to be
observed in everything, for things should not be done
at random, but when there are common sense,
education, and good counsel to guide you, to hesitate
then is to go wrong.
It should not take a man long to decide when there
is a speculation presented him, and his decision should
be obstinately against the speculation. There are too
many good opportunities to succeed in ventures that
are legitimate to touch speculation. It is in the
legitimate field of operations that indecision is so often
fatal.
There is another saying applicable to this subject:
“Be sure you’re right, then go ahead.”
 DON’T BE TOO BIG FOR YOUR
BUSINESS
Most children must creep before they can walk.
The reason is because they are not sure of their small
limbs and try them before venturing to depend upon
them.
When the child can walk he goes right ahead and
walks all his life without fear or hesitation.
It is the same in every line of business. The
business man must know just where he stands all the
time, and he must begin small in order to learn how to
rely upon himself.
You are looking for something big, large,
something you think commensurate with your abilities.
Well, then, let me tell you that you will never find
anything to suit you. You are inflated with your ability,
your importance, and fail to see the small things at
your feet and within your reach that if put together will
aggregate the very big thing you want.
You aim at the moon and feel bad because you do
not hit it. While your aim may be perfectly good and
correct, the object may be too far off for you to hit, or
else you must work yourself within reach of it and then
you will hit it.
Small beginnings have made every great man on
earth. Out of the huts and squalid cabins of the world
have issued men who have conquered the world of
arms and commerce.
You have the advantage of them from an
educational point of view, and think you must be
saddled upon a fiery horse before you know whether
you can ride a steady going one.
The millionaire was not a millionaire when he
started, he was an obscure clerk in a dry goods store
working for wages that you scorn. Reduce your size to
something near the right one and you will see things
differently and take what you can get cheerfully, biding
your time to reach higher. Let your hat fit your head.
 DON’T GET DISCOURAGED
One of the greatest causes for failure in life is
discouragement. It seems to be an element in the life of
every man to be up one day and down the next.
When a man gets up it is possible for him to stay
up by hard work and persistence, but if he permits
himself to go down below his balance he may consider
himself altogether down-and-out.
Failure does not mean that you will not succeed,
because struggle as we may we must meet failure and
look it squarely in the face.
But be not afraid of it, take hold of it by the throat
and compel it to work to your advantage.
The lessons learned during the struggle toward
success, and the ups and downs of the road are
valuable and stand for experience. When a driver has
gone over a hard road once, he knows the rocky
portions and can avoid them when going over it again.
It is human to make mistakes. In fact, it is a
maxim: “It is human to err.”
Knowing this to be inevitable, why repine, or be
discouraged?
Follow the example of the small child who falls
and picks himself up over and over again. By and by,
he can walk without falling down.
Remember this: Every dark cloud has a silver
lining. You see the dark side, but if you make your way
around to the other side you will see the sun shining.
Much of the discouragement is caused by
undertaking more than we can accomplish. If that is
the case, then by leaving off a little here and there we
shall soon reduce our enterprise to a success that we
can handle.
 DON’T BE PREJUDICED
We sometimes dislike a man, or hate him, which is
the same thing, because he possesses certain
peculiarities of person or conduct which are different
from ours, or has ideas that are different from those we
favor.
The man may be a perfect stranger to us, and we
may know nothing about his environments or
conditions under which he lives, or the reasons why he
differs from us—we hate him all the same and take the
other side of the street rather than meet him face to
face.
If we were to look into ourselves we might believe
that this man we dislike, has many reasons for not
liking us.
We show prejudice when we judge any man.
“Judge not, lest ye be judged,” says Christ. You are not
the judge of any man’s conduct, and to judge him
entails slander, backbiting, and conspiracies to his
undoing.
You throw mud at another man. Why? Is it not
because you have some spots yourself and want to
draw attention away from them?
You are afraid that if you boost the other man up
you will lower yourself. Hence you unload upon him
some of your objectionable qualities to lighten your
load.
Every man who does this admits that the other
man is better than he, and hopes by adding his faults to
that other man, to reduce the level to somewhere near
a balance. But experience demonstrates the contrary.